Authors who put the word “first” in their titles generally do so to catch the readers’ (and the editors’) eyes. But sometimes the word really does denote an important milestone, and this is the case for the article in this issue of the Journal by Collins and colleagues (pp. 853–858) (1): this is the first time a pneumococcal vaccine has been shown to reduce colonization in a human challenge study.

Why does this matter?

First, it is upon prevention of colonization, and thus transmission within the population, that the effectiveness of universal pneumococcal conjugate vaccine (PCV) programs primarily depends, even though their design, licensure, and implementation were all based on evidence of their direct protective effects in recipients.

Second, a new generation of pneumococcal vaccines that may overcome the main limitations of the conjugate vaccines (specifically, limited serotype coverage and associated replacement of vaccine with nonvaccine serotypes initially in colonization, and subsequently in disease, as well as high manufacturing complexity and, thus, cost) (2) are well along the pipeline, and an affordable and relatively quick means to assess whether they too may have population-wide indirect effects and, if so, how effectively they might do so would be enormously valuable in guiding prioritization of one candidate over another.

We already know that conjugate vaccines reduce carriage, so this study is less about proving that PCV13 does this (although it does teach us something about how), and more about proving that the experimental challenge approach works. Collins and colleagues do this using an adult pneumococcal nasal colonization challenge model they have developed over several years and have published in this journal (3). The rate of acquisition of type 6B colonization after deliberate challenge was highly significantly lowered to 10% by prior PCV immunization compared with 48% in controls, while those in the vaccinated group who became colonized had fewer bacteria detected in subsequent nasal washes. Both animal studies (4) and human observations (5) have hitherto suggested that the mucosal antibodies induced by conjugate vaccines (6) facilitate killing of bacteria on arrival and prevent acquisition of carriage, although one previous study has suggested that PCVs may also reduce colonization density (7), and indeed it makes good biological sense that one effect should be associated with the other. However, it is the fact that this experimental system is shown to be capable of detecting such differences that really matters, as there are good reasons to suppose that any mucosal effects of new protein-based vaccines may affect clearance of carriage more than prevent acquisition (8).

Similar to all others, this study has its limitations. The challenge protocol involves adults, not young children, and it is among the latter age group, whose immune responses are less mature and experienced, that most pneumococcal carriage, and presumably most acquisition and transmission, occurs. The use of only a single serotype strain does not of itself undermine the proof of principle, but one can speculate that pneumococci prepared in a broth culture for deliberate inoculation may very well not have the same gene expression profile as pneumococci that originate directly from nasal or oral secretions, and this may render successful establishment of colonization in the challenge model less efficient than natural exposure. Finally, there is recent evidence that children vary in their susceptibility to pneumococcal colonization (9), and such factors may also affect adult susceptibility to deliberate colonization as well: not a challenge to the validity of findings in a randomized controlled trial, but another factor contributing to the complexity of pneumococcal ecology.

So with this milestone in the development of the study of human upper respiratory colonization now passed, what are the next steps? Most immediately, of course, will be the application of this challenge model to the study of new pneumococcal vaccines, particularly those containing antigens selected for their potential relevance as inducers of mucosal immune responses (10), and such studies may very well already be in progress. But the other lesson from this study is that we need to stop seeing colonization as a binary endpoint. Because the effectiveness of pneumococcal and other vaccines depends heavily upon their capacity to reduce transmission rates within the population, thus preventing illness among unimmunized people and individuals who fail to make adequate immune responses to vaccine or whose protective responses have waned, as well as those recently and effectively vaccinated, and if vaccines not only prevent acquisition but also suppress the abundance of colonizing microbes, then we have some new questions. To accurately model vaccine indirect effects, we now need to measure variations in colonization density between individuals and age groups and within individuals over time. We need to elucidate the relationships between colonization density and rates of transmission to others and whether and by how much respiratory viral infections change the density and rates of transmission of respiratory tract bacteria (11), and if so, by what mechanisms.

If we really want to make the transition from using inspired guesswork in our vaccine designs and implementation policies, with all the attendant operational and financial risks that come with either underestimating or overestimating effects, then this neglected area of microbiology now needs urgent attention. Collins and colleagues have shown us that it can be done, and done well, in humans. Other colleagues advising grant-giving bodies and journal editors now need to recognize that bacteria do not necessarily need to be causing disease to be interesting and important for medicine. Ultimately, licensing authorities and regulators need to confront the fact that the tried and tested approaches they use, designed for drugs and applied hitherto to vaccines, are not equal to the job. Establishing effectiveness and even safety solely at the level of the individual recipient is ignoring more than half of the story and can only result in vaccines being rejected as inadequate, which could succeed, and even accepting others, which may fail. Without their lead, manufacturers and clinical trialists will struggle to solve this conundrum. New experimental tools including human challenge experiments and cluster randomized studies designed to measure the indirect protective effects of vaccines already exist (12) but are woefully underused. Other tools now need to be developed. Above all, what this report demonstrates beautifully is that empirical challenge studies can be done in humans safely and ethically and can yield solid results using only a fraction of the time and money required for natural history-based trials.