|Home | About | Journals | Submit | Contact Us | Français|
Some screening and treatment programs implemented to control genital Chlamydia trachomatis infections and their complications have shown initial reductions in infection prevalence followed by rises to pre-program levels or higher. One hypothesis is that treatment shortens duration of infection, attenuates development of protective immunity and thereby increases risk of re-infection.
A literature review was undertaken to assess evidence supporting the concept of protective immunity, its characteristics and its laboratory correlates in human chlamydial infection. The discussion is organized around key questions formulated in preparation for the Chlamydia Immunology and Control Expert Advisory Meeting held by the Centers for Disease Control and Prevention in April, 2008.
Definitive human studies are not available, but cross-sectional studies show chlamydia prevalence, organism load and concordance rates in couples decrease with age, and organism load is lower in those with repeat infections, supporting the concept of protective immunity. The protection appears partial and can be overcome upon re-exposure, similar to what is found in rodent models of genital infection. No data are available to define the duration of infection required to confer a degree of immunity or the time course of immunity following resolution of untreated infection. In longitudinal studies of African sex workers, a group presumed to have frequent and ongoing exposure to chlamydial infection, interferon gamma production by peripheral blood mononuclear cells in response to chlamydial heat shock protein 60 was associated with low risk of incident infection. In cross-sectional studies, relevant Th1 type responses are found in infected persons, paralleling the studies in animal models.
The data support the concept that some degree of protective immunity against re-infection develops following human genital infection, although it appears at best partial. It is likely that factors besides population levels of immunity contribute to trends in prevalence observed in screening and treatment programs. Future studies of protective immunity in humans will require longitudinal follow-up of individuals and populations, frequent biological and behavioral sampling and special cohorts to help control for exposure.
Immunity capable of resolving a genital chlamydial infection and protecting against reacquisition of infection has been characterized largely in the mouse and guinea pig animal models, as reviewed in detail by Rank and Whittum-Hudson . To date, it has not been considered safe and practical to experimentally infect humans with C. trachomatis and thus no human challenge studies are available. It is not ethically acceptable to withhold treatment for known infections. Therefore, we cannot define directly in humans the natural course of infection, the course of re-infection at various times after resolution of primary infection or the effect of treatment on development of protective immunity. Because of these limitations, available evidence in humans of protective immunity is mostly indirect and inferred from epidemiologic studies.
Evidence that immunity is capable of resolving infection in humans is found in limited studies of natural history, reviewed in detail by Geisler , which consistently show that humans clear chlamydial genital infections in the absence of antibiotic treatment. Two studies in women suggest that clearance after 1 year is about 45–55% [3, 4] and 94% by 4 years . Another study in pregnancy showed 44% spontaneous clearance over 2–3 months . Two recent studies with much shorter follow-up periods, as well as a prior review, indicate that male genital infections also clear over time [6–8]. The data in aggregate strongly support the concept that immunity capable of resolving infection develops in humans. However, unlike in animal models, the time course to resolution in humans is measured in months to years rather than in weeks. Thus, these same studies clearly demonstrate that untreated infections may persist for many months and suggest C. trachomatis is capable of evading host defenses and/or that an effective human response in some cases is slow to develop.
Our primary focus here is on human protective immunity as measured by reduced detection of infection on re-exposure following a resolved infection, or reduced organism burden or duration of shedding on re-exposure. Defined in this way, protective immunity develops in response to prior infection and might be complete (no detectable infection after re-exposure) or partial, characterized by shorter duration of organism shedding or lower organism burden or both after re-exposure. In animal models of genital infection, complete protection has been observed but is very short-lived, while partial protection is uniformly observed and longer-lasting . We were not able to find solid evidence for this short period of complete protection in human infection. Nevertheless, partial protection is very important from a disease control standpoint, since shorter infection duration and quantitatively less organism shedding are likely associated with decreased risk of transmission. Epidemiologic studies which bear on the concept of protective immunity in the context of genital disease are the focus of our review. The role of immune responses in either promoting or protecting against development of upper tract disease in women is reviewed by Darville and Hiltke .
While useful parallels are found between trachoma and genital infections, important differences in biology and epidemiology exist. Biologically, serovars A and C are strictly associated with trachoma (serovars B and Ba cause trachoma but also cause genital disease), are tropic for conjunctival epithelium and differ from genital strains in that they lack an intact trpA gene and cannot utilize indole to synthesize tryptophan . Other differences are found in genes encoding polymorphic membrane proteins , cytotoxin  and the translocated actin-recruiting phosphoprotein (TARP) , but the relevance of these differences to pathogenesis is currently unclear.
Epidemiologically, trachoma transmission is driven by sharing of ocular secretions among young children within family and community groups, facilitated by poor facial hygiene and the ubiquitous presence of flies . Exposure and re-exposure in hyperendemic areas might be conceived of as more or less continuous. In contrast, exposure to sexually transmitted chlamydial infections is episodic and depends not only on behavior (coitus) but also on the probability that coitus occurs with an infected person. Despite these differences, recent studies related to immunity in trachoma are included in the review when they fill gaps or corroborate data from studies of genital infection. However, trachoma immunity is not the focus of the review and trachoma vaccine studies are not reviewed at all.
Markers of essential mechanisms of immunity defined in animal models, such as chlamydia-specific CD4+ T lymphocytes, interferon gamma and immunoglobulin at mucosal sites, have been documented to exist in humans. These studies are of interest because they provide evidence of responses in humans that are analogous to effective responses defined in model systems, even if the responses have not been studied longitudinally to establish associations between them and protective immunity.
This review is organized around 4 key questions developed by the authors as part of preparation for the Chlamydia Immunology and Control Expert Advisory Meeting April 23–25, 2008. A literature review was undertaken regarding protective immunity in humans, specifically in the context of the contemporary epidemiology of chlamydial genital infections, and implementation and effectiveness of control programs.
An Ovid Medline search was conducted by one of the authors (FX) in January 2008 for articles published since 1950 under the major headings of “immunology”, “immunobiology” or “immune response”; “susceptibility”, “immunity”, “recurrent infections”, “repeat infections”, “subsequent infections”, “re-infections”, or “recurrence”; “T-lymphocytes” or “antibodies”. The search was narrowed to “chlamydia” or “Chlamydia trachomatis”, and to “human” and articles in English. After excluding articles dealing with infants and age >65, a list of titles for 1287 articles were reviewed for relevance. We focused on primary research articles rather than reviews. Of articles evaluating laboratory parameters of immunity, we focused on those evaluating samples obtained from studies of infected persons with or without uninfected controls. We excluded articles focused primarily on in vitro studies using human cell lines. Of the resulting list of 209, two authors (FX and RJ) conducted a second review for relevance and selected 108 for review in detail by all co-authors (RJ, MR, FX, BB). Summaries were prepared and assembled into tables of evidence relating to the key questions. The tables included study design and methods, outcome measures, findings related to key questions, a judgment of strengths and weaknesses and relevance to key questions. Additional articles found through searching reference lists of the identified articles and others found relevant by the panel of reviewers were included. After a summary of the tables was presented at the meeting, the most relevant studies were incorporated into this synopsis.
A limitation of this synopsis is that some informative papers may not have been reviewed. In addition, representative examples of papers presenting certain types of evidence are cited rather than all such articles.
One of the most robust epidemiologic characteristics of both genital and ocular chlamydial infections is higher prevalence among younger versus older people. For example, the highest rates of genital infection are seen in women in the 14–19 year age group and in the 20–29 year age group in men . This inverse relationship between age and prevalence has been interpreted to suggest that protective immunity is acquired over time. However, in young women, the period of highest risk of infection also correlates with the period when cervical ectopy, a risk factor independently related to prevalence of chlamydial infection, is most frequent . In addition, the 18–19 and 20–24 year age groups in both men and women are peak intervals for having ≥2 sex partners in the previous 12 months . A strong inverse relationship of age and prevalence is also noted for gonorrhea  which may be behavior- or developmental-related, since little protective immunity develops in response to gonorrhea because of extensive antigenic variation in gonococci. Therefore, distinguishing the effects of development and behavior versus acquired immunity in chlamydial genital infection is difficult. A large (n=14,605) retrospective cross-sectional culture-based study in high risk men and women in an STD clinic analyzed age in relation to gender, race, history of STD, physical findings, numbers of sexual partners per year and frequency of sex. In multivariate analysis, age was an independent inverse predictor of chlamydia culture positivity in both men and women . Furthermore, when a subset of subjects known to be exposed to chlamydia was analyzed in multivariate analysis, age persisted as an inverse independent predictor of chlamydial infection . However, in this study immune responses as measured by microimmunofluorescence serology and blastogenic responses of peripheral blood mononuclear cells (PBMCs) to chlamydial elementary bodies were not correlated with age.
Similar age-relationships have been identified in trachoma, but in prospective, longitudinal studies. Estimated prevalences and durations of infection, as well as rates of incident infection, are significantly higher in young (0–4 years) versus older (>15 years) children . A more recent study not only confirms the inverse age relationship of trachoma infection and disease, but also suggests that higher levels of local interferon gamma mRNA transcripts are found in older persons . Thus in trachoma, the inverse association of age and C. trachomatis prevalence exists at a site anatomically and physiologically different from the genital tract. This suggests that age-dependent changes in genital tract physiology and sexual behaviors may not be the exclusive reasons for the observed association in epidemiologic studies of genital infection. The results  further associate relevant elements of the mucosal immune response, such as interferon gamma, with age in a manner consistent with acquired protective immunity.
In addition to rates of chlamydial infection, organism load has been inversely related with age, suggesting that acquired immunity may restrict chlamydial replication in older persons. The largest study used quantitative cell culture at the endocervix in high risk women (n=1231) attending an STD clinic . In multivariate analysis, age was inversely correlated with higher inclusion counts. Although there is the potential for confounding with unmeasured developmental events, cervical ectopy was not associated with organism load in univariate analysis.
Several studies have shown an association of higher C. trachomatis infection rates in subjects who self-report no prior history of sexually transmitted disease, suggesting that potentially naïve subjects are more susceptible due to lack of acquired immunity. A study in high-risk men (n=2546) and women (n=1998) attending an STD clinic showed a significantly higher culture isolation rate in those who self-report no STD history . A larger cohort (n=14,605) derived from the same STD clinic also indicated that self-report of no STD history is associated with higher isolation rates . In contrast, other studies show that documented past history of chlamydial infection is associated with increased risk for current infection  and that self-report of past STD is a risk factor for incident infection .
Study of chlamydial infection in sexual partnerships could provide evidence to support the concept of protective immunity, although many factors other than adaptive immunity may affect transmission. The idea here is that low concordance of infection within sexual partnerships indicates the possibility of protective immunity preventing transmission. Quinn et al. used nucleic acid amplification tests (NAATs) to evaluate sexual partnerships in a cross-sectional study in STD clinics . Of 494 partnerships, NAAT was negative in both partners in 79.6%. Of partnerships in which at least one partner was infected, only one partner was infected in 32% and both partners were infected (concordant) in 68%. Young age was a risk factor for concordant infection in both partners . A smaller NAAT- and STD clinic-based study of 106 partnerships found 76% infection concordance . A small contact-tracing study in Boston selected partnerships in which it was reasonable to infer the direction of transmission, in this case from men to women. The concordance rate for NAAT-proven chlamydial infection was 65%. In the same study the concordance rate for Neisseria gonorrhoeae infection, which confers little protective immunity due to extensive antigenic variation, was 73% . Thus, in concordance studies in STD clinic populations, only modest support is found to suggest protective immunity. However, a community-based study in German gynecologic practices comprising 1690 asymptomatic couples screened by NAAT , showed that of 78 couples with at least one positive partner, only 27 (35%) were concordant. The subjects tended to be older and to be in longer relationships than subjects in the STD clinic-based studies. In addition, the rate of concordance decreased with age. Although statistical tests were not reported for this trend, the results are consistent with the possibility of protective immunity acquired with age.
Many studies, mostly in women, have evaluated the timing and risk of repeated infections during a period of observation. These studies demonstrate another robust epidemiologic characteristic of chlamydial genital infections: repeated infections are common. Median time to reoccurrence is typically 5–7 months and risk of incident infections is higher among younger versus older women [29, 30]. A study in school-based health centers reported a cumulative incidence of repeat infections of 26.3% in one year; another in community-based health centers reported that 59.6% with initial infection developed repeat infection on 18 month follow-up . In a study of repeat chlamydial infection sought on scheduled follow-up visits at about 1 and 4 months after treatment of an index infection, cumulative recurrence rate was 13.4% at 4.3 months. The factor associated most strongly with repeat infections was resumption of sexual activity . Finally, a study in sex workers indicates that the strongest epidemiologic predictor of incident infection is C. trachomatis infection at the baseline visit . In aggregate these observations do not exclude the concept of protective immunity, but strongly suggest that it is partial at best.
A few studies have evaluated repeat infections using serotyping or genotyping methods to help define the nature of repeat infection. As a biomarker, strain typing is most helpful when the infecting serovars/genotypes at the 2 episodes are different; this establishes that re-infection is likely responsible for the repeat episode. When the serovars/genotypes at each episode are the same, interpretation is less clear; such episodes could represent re-infection from an untreated partner or antibiotic treatment failure. In STD clinic patients (n=72) repeat infections occurred with the same serovar in 33.8% of cases when the expected rate of same-serovar recurrences based on the distribution of serovars was only 18.4% . In a study of repeat infection in adolescent women, 48 had repeat infections with serovar determination at both episodes over a period of 21 months; 44.8% of the repeat infections were with the same serovar . In both studies it is probable that most of the repeat infections with the same serovar were re-infections, since the majority of subjects had negative cultures between repeat infections. Brunham and colleagues found that in sex workers, 62% (13/21) of repeat infections occurring 1–6 months following an index infection were with the same genotype, while only 11% (2/19) occurring >6 months were the same genotype . The authors suggested that the low proportion of same-genotype infections at >6 months is consistent with serovar-specific immunity, although epidemiologic factors such as partner change were not examined. More recently, a larger study in adolescent women comprising 183 infection pairs with complete genotyping at both episodes also found that early repeat infections were more often than late repeat infections to be with the same genotype; however, the late different genotype infections were significantly associated with partner change . On balance, over the short term, serovar-specific immunity appears not to have a clinically pronounced effect in genital chlamydial infection.
Data from a culture-based study in STD clinic patients suggests that protective immunity may occur but is limited in duration . In both men and women, a laboratory-documented chlamydial infection less than 6 months prior to an index visit was associated with a lower prevalence of infection than a documented chlamydial infection more than 6 months prior. Although suggestive of protective immunity that is limited in duration, it is possible that the more recent treatment in the <6 month groups may have reduced the number of prevalent infections detected at the index visit. Finally, a study using quantitative DNA amplification techniques in a small number of subjects with repeat infections showed that organism load is lower in those with repeat infections, suggesting that prior exposure may restrict replication at the local site as a result of acquired but partial immunity .
Sex workers have frequent sexual interactions and thus the potential for repeated exposure to chlamydia-infected clients is substantial. While studies in sex workers are difficult to generalize to at-risk but non-sex worker populations, the data may allow detection of protective immunity in a setting where sexual activity is consistently high. A series of prospective studies in sex workers in Nairobi have been undertaken to study a highly if not uniformly exposed population in which protective immunity might be detected and its correlates evaluated.
Brunham and colleagues found in multivariate analyses that the probability of incident chlamydial infection was inversely related to duration of prostitution. Further, increased risk of incident infection was associated with HIV status, independent of CD4 count . Both associations suggest that acquired immunity develops in this population and reduces acquisition of incident infection. The same cohort was used to study risk factors for chlamydial pelvic inflammatory disease (PID) . In this analysis, the primary finding related to protective immunity is that among HIV-infected women, those with an entry CD4 count of <400 were at highest risk for developing PID, suggesting that CD4 lymphocytes are required to restrict chlamydial replication and prevent establishment of upper tract disease. This finding was followed-up in a separate cohort enrolled based on clinical diagnosis of PID, where peripheral blood mononuclear cells (PBMCs) from a subset of 95 women were evaluated for cytokine production in response to stimulation by whole elementary body antigens. Here, interferon gamma production was lower in HIV-infected subjects compared to HIV-uninfected subjects; further, the responses are associated with lower CD4 lymphocyte counts . These laboratory evaluations are consistent with the clinical correlations and suggest that Th1 mechanisms are required to limit upper tract pathology.
Cohen et al. reported incident chlamydial infections as a function of both epidemiologic and laboratory measures of immune responses in a cohort formed to study the immunoepidemiology of STIs in female sex workers . In this study, 299 women were enrolled and followed prospectively for incident chlamydial infection. The major epidemiologic factors associated with risk of incident chlamydial infection included C. trachomatis infection at enrollment, incident Neisseria gonorrhoeae infection, young age and less than 2 years duration of sex work. In contrast to the earlier study , HIV-1 infection did not correlate with risk of incident chlamydial infection. The inverse association of young age and fewer years of sex work and incident infection are consistent with the concept of protective immunity. However, baseline C. trachomatis infection and incident gonococcal infections were strong risk factors for incident infection. These results suggest that these women were exposed to a subset of men with a higher prevalence of C. trachomatis and that protective immunity is partial, although the prevalence of C. trachomatis infections among the clients of the sex workers was not reported. The major laboratory correlate of decreased risk of incident infection was interferon-gamma production by PBMCs stimulated by recombinant chlamydial heat shock protein 60 (cHSP60) at baseline. Among 29 subjects positive by this measure, no incident chlamydial infections were observed. The importance of this specific laboratory measure is supported by the results of another study where PID and history of repeated chlamydial infection were associated with reduced interferon gamma production in PBMCs stimulated by cHSP60 .
Several lines of evidence indicate that protective immunity to re-infection develops in humans over time. Protective immunity is likely to be partial at best and can be overcome on re-exposure, paralleling the experience in animal models. To the extent that data are available, there is little convincing clinical evidence that immunity is serovar/ompA genotype-specific. The duration and strength of partial protective immunity in humans is uncertain. Prospective studies in sex workers and trachoma suggest that Th1 type immune mechanisms including CD4 lymphocytes and interferon gamma are important components of acquired immunity in humans. Although definitive human studies are not available, the picture of human protective immunity that emerges from the literature correlates well with characteristics and mechanisms of partial protection defined in mouse and guinea pig models of genital infection; the short-term complete immunity early after resolution of infection observed in the rodent models has not been demonstrated in human studies.
The primary measures of partial immunity in animal models are diminished intensity of shedding (organism load) and decreased duration of infection. Human studies using chlamydia incidence as an endpoint are not fully capable of detecting and characterizing such partial immunity. To further characterize protective responses in humans, longitudinal studies are required that include baseline and subsequent measurement of identified candidate markers such as interferon gamma production by cHSP60 stimulation of PBMCs, together with serial sampling for incident infection and determination of organism load. Scheduled periods of frequent non-invasive sampling performed prospectively may identify incident infection “in real time” [42, 43] and perhaps estimate its duration. Study of special cohorts, for example those including sexual partnerships (dyads), may provide an opportunity to assess candidate markers and their ability to predict incident infection given known exposure.
Most of the repeat infection studies discussed above share the characteristic that index infections, of variable but mostly unknown duration, were treated and thus the natural course of infection truncated. This raises the question as to whether treatment of first or repeated chlamydial genital infections hampers the development of protective immunity that might ordinarily develop if a long-lasting infection were to go to spontaneous resolution.
A study in the mouse model of genital infection with C. muridarum clearly shows that antibiotic treatment, given before immune responses are fully developed, attenuates the development of protective immunity . Doxycycline, begun at 4 different times after inoculation for a primary infection (day 0, day 3, day 7 and day 10), was given for 14 days. Treatment promptly terminated chlamydial shedding and prevented or reduced the frequency of hydrosalpinx. The course of infection following rechallenge was longer with higher organism shedding in all of the treatment groups compared to animals allowed to clear the primary infection without treatment; the earlier treatment was started, the more profound the effect. Measures of immune response (local IgA, serum IgG and interferon gamma production by chlamydia-stimulated splenocytes) were correspondingly reduced.
It seems likely that a similar phenomenon would be observed if experiments of this kind were possible in humans. We know that chlamydia-specific responses of peripheral blood mononuclear cells , infection -associated endocervical T cell infiltrate  and presence of IL-12  at the endocervix decline or resolve within 3–4 weeks after antibiotic treatment of genital infection. The major unknown is how long humans must be infected before partial protection develops. In mice, it is a matter of just a few weeks. In people, prevalent infections can have durations of a few weeks to several years depending on when or if they are detected. Duration of incident infections depends on the frequency of screening and treatment. To our knowledge, there is no direct evidence that defines the minimum duration of infection that confers a given degree of protection. As can be seen from the mouse experiments, longer duration of infection produces protective immunity but also allows the development of upper tract pathology in the majority of animals . Although to a substantially lesser degree , the same is likely true in humans [6, 48].
Two epidemiological studies suggest that treatment for C. trachomatis infection is associated with increased rates of re-infection. In each case, the authors hypothesize that treatment hinders the development of protective immunity. Two additional studies show that prevalence of serum antibodies to C. trachomatis are falling during a period when case rates and seroincidence are rising.
The first study evaluated repeat genital infections in the Vancouver metropolitan area. This large (33,917 infected individuals) study was undertaken because reports of chlamydial infections that had initially decreased after initiation of a screening and treatment program rose again after a number of years . The hypothesis that program-related screening and treatment has increased population susceptibility depends on an overall decreasing interval between infection acquisition and treatment. However, neither the duration of infections prior to treatment nor the interval between positive test and treatment was reported. Nevertheless, increasing relative rates of re-infections were observed from 1989 to 2003 with increased risk among younger individuals and women. Low rate of re-infections early in the observation period are expected because it takes time for subsequent re- infections to accrue; it is not surprising that as more and more individuals previously screened are re-screened, re-infections increase as a proportion of total infections. A transmission model is presented based on a set of assumptions regarding protective immunity and ranges of various parameters related to the duration of infection as a function of prior infection. A modeled control strategy which shortens duration of infection (for example, by prompt treatment) recapitulates the empirical observations. The study results are consistent with the hypothesis of a population phenomenon of reduced immunity, although alternative reasons for the observations have been suggested based on limitations of population surveillance and screening data. These include: increased screening coverage ; increased use of NAATs for diagnosis ; changes in proportions of high- and low-risk individuals screened, testing frequency and uncertainty in establishing re-infection rates ; and contemporaneous factors such as behavior change related to the HIV epidemic  .
The second study reports rates of re-infection following targeted azithromycin treatment in the context of trachoma . In the analysis of re-infection, subjects with NAAT-documented C. trachomatis at baseline were identified in trachoma-endemic Vietnamese communes; the duration of infection prior to baseline testing was not reported. Two communes were given targeted treatment with azithromycin at baseline and 12 months. Treatment was not based on infection status, which was determined post hoc, but was given to those with clinically identified active trachoma and their household contacts. The targeted treatment strategy resulted in treatment of ≤ 11% of infected individuals in these communes, indicating a substantial pool of infected persons remained. In the third commune no systemic antibiotics were given. Re-infection rates were examined at 18, 24 and 36 months. At 36 months, significantly higher re-infection rates were observed among individuals in the communes where azithromycin was given, although the total number of individuals infected at baseline who actually received azithromycin appears small. The authors attribute the higher rates of re-infection to impaired development of immunity.
A third study is based on the seroprevalence of IgG antibodies against C. trachomatis in Finland . Here, sera were analyzed from 8000 women representing a subset of the Finnish Maternity Cohort serum bank, stratified by calendar year and by age of the women at the time of sampling. Seroprevalence was measured in samples obtained between 1983 and 2003. In women under 23 years of age, seroprevalence fell from 16.0% in the 1990–1996 interval to 10.6% in the 1997–2003 interval. A similar fall was observed in women 23–28 years of age in the same time periods (19.1% to 12.5%). The authors note that the number of reports of chlamydial infections has risen in recent years; specific data available from the Finnish Statistical Database of the Infectious Diseases Register (http://www3.ktl.fi/) show an increase from 8031 cases in 1995 to 12,863 in 2003. An increase in case rates coupled with a decrease in seroprevalence is consistent with a population decrease in protective immunity, although many other factors could contribute to an increase in reported case rates, including increased coverage and frequency of screening.
Finally, a recent follow-up study using the Finnish Maternity Cohort shows that the rate of seroconversion was higher in the 2001–2003 3-year time period compared to the 1983–1985 period in the group of women 23–28 years of age (OR 3.2, 95% CI 1.1–8.7) . However, considering all 3 year periods over the entire time period, there was not a significant time trend in this age group (p=0.10). There were no significant time trends in seroconversion in women under 23 years of age. An increase in seroincidence coupled with a decrease in seroprevalence would be consistent with a population decrease in immunity; however, the reported seroincidence data are at present inconclusive.
No studies are available that establish the minimum duration of chlamydial infection in humans required to elicit protective immune responses. Likewise, no direct data are available to indicate the point during the course of human infection at which antibiotic therapy can hinder development of relevant protective responses. In the animal study, early treatment appears required (within 10 days) but given the large difference in duration of mouse versus human genital infections, it is difficult to extrapolate the results in the model to human infection. Some measures of cell mediated responses in humans diminish shortly after antibiotic treatment. Epidemiologic studies of re-infection rates, chlamydia case report trends, and seroprevalence and seroincidence trends suggest that intensified case finding and treatment in some locales are associated with an initial decrease followed by paradoxical increases in reported chlamydial cases and perhaps seroincidence. The fall in seroprevalence rates in Finland provides initial evidence that conveniently measured immune responses at the population level are less prevalent during an interval with an increase in reported cases. A reasonable hypothesis is that prompt treatment blunts the protective response that in earlier pre-program eras developed after long periods of untreated infection. A study in mice establishes proof-of-principle in a model system. However, further study is needed to better understand the relative roles of other potential explanations in these prevalence and incidence trends and re-infection rates by characterizing screening coverage, frequency of screening and other relevant epidemiological variables. In addition, longitudinal study of serologies in selected cohorts may provide insight into duration of serum antibodies and their relationship to incident chlamydial infection.
Based on data available from studies in animal models , the key elements of resolving and protective immunity include: trafficking of chlamydia-specific CD4 lymphocytes to the genital site; production of Th1-related cytokines including interferon gamma capable of restricting chlamydial growth and presence of neutralizing IgG antibody at the local site. From the animal models, it is clear that the number of CD4 lymphocytes at the mucosal site decreases as infection is cleared and this reduction is correlated with the return of partial susceptibility to re-infection. On re-challenge, local antibodies likely reduce the infectivity of any given inoculum by neutralizing some proportion of elementary bodies, restricting the early intensity of shedding. Rapid anamnestic recruitment of memory T cells to the mucosal site more rapidly clears those re-infections. Together these mechanisms account for the observed partial immunity. Here we review representative studies to establish that elements of these essential responses are observed in human infection. The studies reviewed in this section are mostly cross-sectional and thus in most instances the measured responses cannot be directly linked to protective immunity except as noted.
An early study reported that 1) interferon gamma can be found in endocervical secretions; 2) endocervical levels were higher in chlamydia-infected versus uninfected women; and 3) plasma levels did not correlate with infection. The major significance of this work was to document an important Th1 cytokine at the site of infection in humans . A more recent study suggested that higher endocervical levels of interferon gamma were seen in women with recurrent compared to primary infection . In a multivariate analysis of endocervical cytokines and clinical characteristics, interferon gamma, IL-12 and IL-10 were significantly associated with endocervical infection with C. trachomatis (n=17) . In a study of 396 female adolescents, endocervical secretions were assayed for IL-2 and IL-12; infected women had lower levels of IL-2 and higher levels of IL-12 than uninfected women. In addition, these relationships were confirmed in 96 women who contributed paired samples, documenting the changes as women moved from uninfected to infected state or vice versa . IL-12 is produced by dendritic cells and induces interferon gamma production by T cells, biasing toward a Th1 response.
A careful immunohistochemical evaluation of the human vagina and cervix established that T-lymphocyte subsets, antigen-presenting cells, macrophages and dendritic cells are most abundant in the endocervical transition zone , suggesting that the endocervix, the primary site for lower genital tract chlamydial infection, is also the major inductive/effector site for cell mediated immunity in the lower genital tract. The data show that non-invasive sampling via cytobrush or other cell-collection techniques is useful in evaluating local responses to chlamydial infection. In a flow cytometry study of endocervical cells, CD4+, CD8+ and CD83+ (dendritic cell) phenotypes were more frequent in infected versus uninfected women, while CD19+ (B cell) phenotypes were not different. There also was no difference in peripheral blood mononuclear cell phenotypes between infected and uninfected women .
In a recent carefully conducted study, 20 infected women were sampled by cytobrush at the endocervix both at the time of infection and 1 month following antibiotic therapy . At the time of infection, accumulation of neutrophils and of CD4 and CD8 T lymphocytes was observed. Both during infection and after treatment, the predominant endocervical cell infiltrate was CD45RO expressing effector memory T cells. Endocervical T cells also expressed CD103 consistent with mucosal homing. HLA-DR expression by T cells was significantly increased during infection indicating activation. After treatment and documented clearance of chlamydiae, CD3+ cells were markedly reduced. There were no differences in peripheral blood mononuclear cell phenotypes between infected and uninfected women with the exception of a higher proportion of CCR5+ T lymphocytes after treatment.
In summary, these studies indicate that local Th1 cytokines, mainly interferon gamma, are associated with C. trachomatis infection and an infiltrate of T lymphocytes predominates at the local site. The T lymphocyte infiltration is largely cleared within 1 month following treatment .These findings are consistent with the changes associated with infection and clearance in both the mouse and guinea pig models.
Local antibodies derived from persons with trachoma are capable of neutralizing chlamydial infectivity in animal models . In addition, presence of antibody in local secretions, particularly IgA, was found to correlate inversely with quantitative culture from the endocervix . Cross sectional studies have noted high rates of serum IgG antibodies to whole elementary bodies among STD clinic attendees, but in one large study serologies were not associated with age or reduced isolation rates of C. trachomatis .
While interferon gamma production by PBMCs in response to cHSP60 was significantly associated with protection against incident chlamydial infection in sex workers, endocervical IgG and IgA to whole elementary bodies or cHSP60 were not . In addition, serum IgG antibodies to whole elementary bodies and cHSP60 were not associated with a lower risk of incident chlamydial infections.
Many contemporary studies of antibody bear on risk of sequelae of chlamydial infections such as PID , PID recurrence and lower pregnancy rates . Relationships between chlamydial serology and upper tract sequelae are reviewed elsewhere in this supplement by Haggerty et al.  and Darville and Hiltke .
Serological responses have been documented to other chlamydial proteins as measured by immunoblotting  and to specific proteins such as plasmid protein pgp3  and chlamydial proteasome/protease-like activity factor (CPAF) . However, these responses have not been examined in longitudinal studies in relation to incident chlamydial infection.
In summary, antibody responses, including those measured in endocervical secretions, have not been found to correlate with protective immunity, but appear to be markers of prior infection. Certain antibodies to cHSP60 or high titers of antibodies to whole elementary bodies appear to correlate with sequelae. Major unanswered questions include how long serum IgG antibodies persist after an episode of infection; whether disappearance of such antibodies is a marker of increased susceptibility to infection and whether early treatment of infection prevents development of a serological response or shortens the period during which antibodies are detectable.
Although mucosal cellular responses are most relevant given the epithelial location of infection, careful studies of systemic cellular responses have been made and are reviewed briefly . A notable series of papers by the DeMars laboratory have defined HLA Class I and II presented T cell epitopes in the major outer membrane protein of C. trachomatis in STD clinic attendees with genital chlamydial infection. MOMP epitopes that activate HLA Class II-restricted T cells from humans with genital infections have been defined . Human genital tract infections were also found to induce HLA Class I-restricted CD8+ cytotoxic T-lymphocytes specific for MOMP . Most of the epitopes were found in constant sequence regions of the MOMP and were species-specific. Subsequently, 5 MOMP peptides represented in variable sequence regions (where serovar-specific B cell epitopes reside) were found to contain HLA Class II presented T cell epitopes that are serovar-specific. Finally, these authors used HLA-A2 tetramers to characterize MOMP-specific cytotoxic T lymphocyte responses and found MOMP-specific T cells in peripheral blood of infected but not uninfected persons . These findings were then applied to a clinical study of trachoma where the frequency of MOMP-specific CD8+ T lymphocytes in peripheral blood were sought. Frequency of these cells correlated with active ocular infection, but sample sizes were too small to relate to protection or disease outcome .
At this time, the only documented marker of protective immunity was that established by Cohen et al . Specifically, interferon-gamma production by PBMCs stimulated by cHSP60 was found in a subset of 29 women in a longitudinal study of chlamydial incidence in Nairobi sex workers. None of these women acquired incident chlamydial infection over a study period of 24 months, a significantly lower rate than seen in women without such responses. While a larger group of subjects had production of interferon-gamma by PBMCs stimulated by whole elementary bodies, the rates of incident chlamydial infections were similar to women without such responses. It is possible that the cHSP60 responses in a subset of these women were masked by responses to non-protective T cell antigens represented on whole elementary bodies.
Nevertheless, the finding that PBMC production of interferon gamma in response to cHSP60 stimulation is associated with decreased risk of incident infection is also consistent with findings in sex workers with PID. HIV positive sex workers, known to have an increased risk of chlamydial PID over women who are HIV negative, have a lower frequency of interferon gamma production in response to stimulation with C. trachomatis antigens . In the latter study, PBMCs were stimulated with either whole elementary bodies or MOMP isolated by detergent extraction from elementary bodies. Future longitudinal studies in humans should clearly include similar assays using set of these antigens (whole elementary bodies, MOMP, cHSP60) to confirm the associations found in populations of African sex workers.
Several articles examine HLA alleles in the context of PID associated with chlamydial infection [39, 75, 76], are reviewed by Darville and Hilke  and not further considered here. Two articles summarize selected host factors in adolescents enrolled in the Reaching for Excellence in Adolescent Care and Health study, a longitudinal study with periodic biological sampling . Similar to other studies, young age, multiple partners and prior chlamydial infection predicted incident C. trachomatis infection on follow-up. In addition, HLA class II allele DQB1*06 and HLA class I haplotype B*44-Cw*04 were associated with incident chlamydial infection . The second article focuses on recurrent infections in a subset of 90 adolescents with repeat documented chlamydial infections separated by a negative assay . In multivariate analyses, HLA class II variants DRB1*03-DQB1*04 and DQB1*06 were associated with recurrence, while IL-10 promoter variants were underrepresented in those with recurrent infections.
Other than genetic factors, associations have been reported between chlamydial infection, quantitative chlamydial shedding at the endocervix and concomitant gonococcal infection [21, 34, 36]. The basis for this association is not clear, although gonococcal IgA 1 protease has been suggested as a biological factor . In a cross-sectional study, bacterial vaginosis has been identified as a risk factor for acquisition of chlamydial and gonococcal infections in women recently exposed to a partner with urethritis . Women harboring hydrogen peroxide producing lactobacilli were less likely to be infected by either organism. Since the altered microbiota of bacterial vaginosis frequently include indole-producing anaerobes, some have hypothesized that the increased risk of chlamydial infection in bacterial vaginosis is related to genital strains’ ability to synthesize tryptophan from indole, thereby in part evading the activity of interferon gamma at the infected site .
The primary organism factor that has been evaluated in the context of human infection is serovar determination either by immunoassay in older studies and more recently direct sequencing of the ompA gene from clinical samples. Serotyping/genotyping studies performed in the context of repeated human infections are reviewed above [34–36]. Other serotyping/genotyping studies have been focused largely on establishing strain distributions in various populations and locales and associating clinical phenotypes with ompA polymorphisms. Studies in sex workers indicate considerable strain variability found in variable sequence regions [36, 82]. Overall, serovars E, D, F and Ia are seen most frequently with some regional variations in the United States [83–85]; similar distributions are also noted in Europe, Asia and Australia [86–89]. The classic disease-causing groups of C. trachomatis (trachoma, genital, lymphogranuloma venereum strains) and the associated tissue tropisms do not correlate with serovar/ompA genotype [83, 85]; however, sequence variations in some polymorphic membrane protein genes do . Newer multilocus genotyping methods have been described but clinical correlations are limited to date [90–93].
There are many obstacles in human immunology research. Unlike in animal models, human populations are genetically diverse and even a carefully selected cohort will likely include a heterogeneous collection of those who have little or no protective immunity, various degrees of partial immunity and various levels of exposure to infected partners. However, our synopsis supports the view that a degree of partial protective immunity develops as a result of genital chlamydial infection. In addition, the data suggests that interferon gamma production by PBMCs in response to cHSP60 represents a phenotypic marker of protective immunity for future studies. Our review also confirms the view that longitudinal studies of persons at risk with serial collection of behavioral and biological data are necessary to further define elements of protective immunity and the responsible mechanisms. In the era of programs for screening and treatment of at risk young people, there is concern that prompt diagnosis and treatment increases risk of subsequent re-infection. It is not clear whether the observed increases in re-infections are related to changes in levels of immunity within the populations studied or related to epidemiologic and ascertainment factors related to screening and treatment. Additional research on the role of protective immunity in explaining these findings could help to optimize current strategies until effective vaccines for primary prevention of chlamydial infections are available.
We thank Sami Gottlieb and Stanley Spinola for helpful discussion and critical review of the manuscript.
Financial support: B.E.B. supported by NIH NIAID Sexually Transmitted Infections Cooperative Research Center U19AI031494 (Stanley Spinola)
Potential conflicts of interest: none reported
Presented in part: Chlamydia Immunology and Control Expert Advisory Meeting, Atlanta, Georgia, 23–25 April 2008.