Challenge experiments have been an important method of studying the pathogenesis of many infectious diseases and of evaluating initial efficacy of vaccines before large scale field tests are conducted.^1,2 In challenge experiments, infections are deliberately induced under carefully controlled and monitored conditions to healthy research volunteers. Induced infections are usually either self limiting or can be fully treated within a short period of time. Because physicians should be dedicated to alleviating disease and avoiding harm to patients, this type of experiment may cause uncomfortable symptoms and evoke serious moral concerns. It should be appreciated however that clinical research commonly involves risks to subjects that are not outweighed by medical benefits but are justified by the potential to acquire new knowledge.³ In that regard, infection inducing challenges are not necessarily more ethically problematic than phase I trials aimed at determining maximum tolerated doses of medications. Like any clinical research, challenge experiments should be conducted by competent investigators according to sound protocols that incorporate appropriate safeguards to ensure the safety of volunteers. Because these experiments may provide valuable information that might not be otherwise obtained, lead to novel therapies, or speed up vaccine development that will ultimately spare morbidity or death from infectious diseases and reduce exposure of large groups in field trials, challenge experiments may be justified.⁴ However, the scientific rationale should be carefully examined for any given pathogen and model. When such a rationale exists, then the question of risks and discomforts should be addressed.

Twenty years after the first culture of Helicobacter pylori there are still many gaps in our knowledge of this worldwide infection. In particular, little information is available regarding the precise mode of transmission of the infection, the conditions required for exposure to H pylori to lead to chronic gastric infection, and the early clinical and pathological events following colonisation of the gastric mucosa.⁵ Indeed, the infection is typically acquired in childhood but most studies related to H pylori pathogenesis and host response are conducted, for obvious ethical and practical reasons, in adults. As young children are not always chronically infected following initial infection,⁶ a better understanding of the early phase of the infection may prove useful in the development of novel therapies and vaccines. Although it is not entirely clear that the early events in H pylori infection in adults are identical to those in children during natural infection, adult infection has been reported and has led to typical chronic infection.⁷ A H pylori challenge model in adults may thus present an unique opportunity to obtain information on the pathogenesis of this bacterium, which may be potentially useful in the development of new drugs or obtaining knowledge on the host response crucial for vaccine development.

By inducing experimental H pylori infection in healthy volunteers, the report of Graham and colleagues⁸ in this issue of Gut has certainly contributed to our understanding of this infection (see page 1235). This study provides valuable information on the low inoculum size required for infection, the impact of some virulence factors, the symptoms associated with acute H pylori infection, the physiological changes induced by H pylori, as well as on the rapid development of histological changes previously thought to be linked to chronic infection. This harvest of valuable information would have been very difficult to obtain by other means, especially if early infection is to be analysed in children. These data represent a well controlled body of information, useful to the scientific community, to better understand transmission and to further validate animal models for pathogenesis studies.

In addition to its use in studying the pathogenesis of infectious diseases, infection inducing challenge experiments have been used to evaluate the initial efficacy of vaccines before conducting large scale field tests for many infectious diseases, including enteric pathogens.⁹ Typically, this step is undertaken after basic research has provided data regarding potential protective antigens, and allowed for a description of the host immune response. Then, ideally, animal models that mimic human infection and response are used to test efficacy before human studies are considered. Finally, candidate vaccine preparations should then be evaluated for safety and immunogenicity in humans, outside of the challenge setting, to minimise exposure of volunteers only to the most promising candidates. In this instance, however, it is of central importance to determine whether the proposed human challenge model is not only suited to reproduce the natural infection but also corresponds to the population that the vaccine is intended to protect.

Vaccine development against H pylori started over 10 years ago, after a proof of principle was established in mice infected with Helicobacter felis.^10,11 In this model, and later in H pylori mouse models, several protective antigens of H pylori were identified. Urease was also shown to be effective in a vaccine preparation administered to infected animals.¹² Urease, cytotoxin associated gene A (CagA), vacuolating cytotoxin (VacA), and H pylori neutrophil activating protein (HpNAP) were also shown to be safe and immunogenic in humans once their protective potential had been established in mice.¹³ Although these antigens are protective in mice, and elicit both humoral and cellular immune responses, no clear immune correlate of protection could be identified in mice, which prevents the use of an immunological test to evaluate the protective potential of a candidate vaccine in humans.¹⁴ Field trials thus represent the alternative, with protection as the primary outcome. As H pylori infection is acquired early in life, this would imply that field trials would have to be conducted in children, in areas of high incidence (that is, on the less favoured children of most societies). If a prophylactic vaccine is to be developed, such a field approach would certainly elicit several serious ethical concerns. These concerns may be progressively addressed if benefit can be shown for children but the incentive to invest in the long research needed to follow that approach would certainly be helped by positive results in adults. As natural infection is rare in adults, a challenge model would then be required to progress substantially towards prophylactic vaccine development. In this context, the model developed by Graham and colleagues⁸ certainly represents an advance. The value of this adult model in predicting vaccine induced protection in children however would further depend on the assumptions that both infection and vaccine induced protective responses are similar in adults and children. In addition, the model would not be appropriate for testing the protective potential of CagA, as the strain was selected as CagA negative for safety reasons. As virulence factors often represent important vaccine antigens, this is certainly a limitation of the model which could be resolved only after extensive safety data are obtained with the challenge model. When the goal is ultimately to develop a therapeutic vaccine however, testing could be conducted by vaccinating adults naturally infected with H pylori.¹⁵ As far as vaccine development against H pylori is considered, it is not yet clearly established that an infection inducing challenge model is likely to facilitate or speed up vaccine development.

In exploring the ethical justification of particular infection inducing challenge experiments, the nature of the infection has to be considered. Legitimate challenge experiments may include infection likely to produce mild symptoms not interfering with the subject’s daily activities and those that are self limiting or that can be adequately eradicated with certainty.³ From the reports of the two investigators that infected themselves with H pylori and from iatrogenic infection reports, it was not expected that acute H pylori would cause more than mild dyspeptic symptoms. To further minimise this risk, the strain was obtained from a patient with no alarming symptom, histological changes, or complications of infection. The primary concern however was that experimental H pylori infection may become chronic, and that a treatment regimen with 100% effectiveness to eradicate the infection was not available.¹⁶ Careful attention was paid by the investigators to selection of the challenge strain. It was derived from a patient that was eradicated without difficulty, and the strain tested for its lack of resistance to the antibiotics commonly used to treat the infection. However, it was CagA negative and the lack of this virulence factor had been associated with decreased eradication rates. Despite this, full confidence that H pylori infection, even with a fully antibiotic sensitive strain, can be eradicated is lacking. Volunteers that might remain infected would then be at increased risk, albeit a small risk, of morbidity. In addition, as the precise transmission mode of H pylori is as yet unclear, there was a small risk that the experimental strain would pass to other individuals. In the course of their study, Graham et al achieved 100% eradication rate, and no apparent transmission of the infection was observed, possibly related to the precautions undertaken to minimise this risk. The size of the experiment however was far too small to ascertain that the strain can be safely eradicated and is not transmitted, with type II errors for these observations being possible. Therefore, the challenge model can not be considered safe, but as progress is made towards the control of H pylori infection, the threshold for acceptability of this model, as for other infections, is likely to shift.

After the experiment had started, an expert panel assembled on 30 November 2000 by the National Institute for Health, Division of Microbiology and Infectious Diseases, concluded that the development of the H pylori infection challenge model should not be pursued at this time.¹⁷ In brief, the reasons for this conclusion were that (1) there was no compelling evidence for the need for a Helicobacter vaccine in the USA, (2) no clear vaccination strategy was defined for the USA, (3) volunteers were at risk of unsuccessful eradication with possible late consequences, (4) animals models had not been fully investigated for vaccine efficacy, and (5) more conventional field trials were available to measure vaccine efficacy. These conclusions however may underestimate the full extent of the H pylori problem worldwide and do not address the scientific rationale for the use of a challenge model for Helicobacter vaccine development. The H pylori challenge experiment reported by Graham et al was conducted under the best available conditions and resulted in minimal harm to volunteers. The results shed light on interesting aspects of the pathogenesis of the bacteria, and provide validation information for the development of animal models. However, there are many more questions to be answered before this model can be validated and proven useful for vaccine development. The scientific community therefore needs to think carefully before deciding to use this model in humans for the development of a vaccine against H pylori.