Introduction

Chlamydia trachomatis (Ct) infection, if untreated in women, can result in pelvic inflammatory disease, a condition leading to significant reproductive morbidity [1–3]. Opportunistic or screening programmes have been recommended or implemented in several countries to reduce prevalence and, subsequently, incidence and reproductive sequelae [4, 5], but their effectiveness has never been confirmed by randomised controlled trials. While findings from the United Kingdom (UK) screening programme (aimed at all those under 25 years) provide a measure of current prevalence of those tested, a declining cumulative risk of infection would be a better marker of success [6–8].

We previously produced an indirect lgG Enzyme Linked Immunosorbent Assay (ELISA) to detect antibody to Ct-specific Pgp3 protein [9]. The Pgp3 protein is transcribed from the highly conserved Ct plasmid [10] that is not found in human C. pneumoniae isolates [11]. Pgp3 is also highly immunogenic in its native, trimeric form [12, 13] and antibody to Pgp3 does not cross react with C. pneumoniae proteins with which Ct shares many similar genes [8]. We have demonstrated the Pgp3 indirect ELISA is significantly more sensitive in detecting past Ct infection than three of the most commonly used ELISAs [9]. We refer to sensitivity as the proportion of individuals with past infection identified as positive by the assay i.e diagnostic sensitivity [14]. To maximise detection rates of past infection, we have developed a Pgp3 double-antigen sandwich ELISA, as has been done for other infections, including HIV-1, Hepatitis B and Hepatitis E [15–17]. Double-antigen assays require that each protein-specific antibody recognises the specific epitope of the antigen bound to an ELISA plate, as well as binding the same epitope on labelled Pgp3, allowing detection of lower antibody titres [18].

In this report we describe the new ELISA, and demonstrate its enhanced performance over other assays using the same samples that were originally used to validate our indirect assay [9]. We then examined Pgp3 antibody associations in participants of the New Zealand Dunedin Multidisciplinary Health and Development Study (DMHDS), a birth cohort study in which detailed information on sexual behaviour and health have been collected at regular intervals from age 18–38 years [19]. Stored sera collected at ages 26, 32 and 38 years were tested by our double-antigen ELISA, the findings compared with self-reported Ct infection and sexual behaviour, and the persistence of the antibody response measured over this 12-year period were determined.

Methods

Results

Pgp3 double-antigen ELISA

By ROC analysis, the Pgp3 double-antigen ELISA gives a significantly higher AUC than all other assays assessed (p<0.0001–0.0002) (Table 1), facilitating discrimination between positive and negative sera. This is confirmed by pairwise comparisons of sensitivities of the assays (Table 2 and Tables A and B in S1 Text). A marked gender difference in the sensitivity of the double-antigen ELISA was observed. Of the 158 GUM clinic samples from women, 131 (82.9% [95% CI 77.0–88.8%]) were Ct-antibody positive (Table 2). This constitutes a 15.9% increase in sensitivity over the indirect ELISA (71.5% [95% CI 64.5–78.6%]). In men, sensitivity was 54.4% (95% CI 47.2–61.6%) (Table 2), a 15.0% increase over the indirect assay. In the 69 men with only one episode of Ct and no history of urethritis the sensitivity was 42.6% (95% CI 31.6–54.5%).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Comparison of ROC analysis AUCs across the different assays.

https://doi.org/10.1371/journal.pone.0151497.t001

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Sensitivity of the Ct antibody assays according to gender.

https://doi.org/10.1371/journal.pone.0151497.t002

Specificity of the double-antigen ELISA, determined on 494 Ct-negative paediatric sera, was 97.8% (95% CI 96.5–99.1%), the same as that for the indirect assay 97.8% (95% CI 96.5–99.1%), and comparable with the commercial Major Outer Membrane Protein (MOMP) assays (95% CI 94.7–99.0%) (Table 3 and Table C in S1 text).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Specificities of the Ct antibody assays using true negative Ct control sera^a

https://doi.org/10.1371/journal.pone.0151497.t003

Ct-specific Pgp3 antibodies in Dunedin Study samples

All Dunedin samples (n = 2641), including 10 for which there was no corresponding behavioural information, were assayed by the double-antigen and, for direct comparison, the indirect ELISAs (Table D in S1 Text). Of the 2365 samples that tested negative by the indirect, 243 were positive by the double-sandwich ELISA, consistent with the higher sensitivity of the new assay. Of the samples that tested positive by the indirect (n = 276), 260 (94.2%) remained positive by the double-antigen ELISA. Of the 16 discordant samples, 11 (from six individuals) were positive when assayed on an indirect ELISA without antigen present, hence those must have been false-positives. The remaining five samples (from three individuals) originally positive were negative by the second assay. This remains unexplained, although non-specific binding due to heterophilic antibodies and rheumatoid factor is a common occurrence in serological assays [29]. The double-antigen ELISA detects nearly double (1.75 times) the number of self-reported Ct cases compared to the indirect ELISA. Hence, only results from the double-antigen ELISA were used for the following analysis.

At all ages, more women than men were Pgp3 antibody positive (Tables 4 and 5). The proportion of seropositive women and men were, respectively, 24.1% and 10.7% (p<0.001) at age 26, 26.2% and 14.0% (p<0.001) at age 32, and 26.8% and 13.1% (p<0.001) at age 38. Among women who self-reported Ct, the percentages seropositive at ages 26, 32 and 38 were 79.5%, 75.0% and 74.6%, respectively; all significantly (p<0.001) higher than those who had never reported infection (Table 4). Among men self-reporting Ct by these ages, 25.0%, 33.3% and 27.0% were seropositive, respectively, again significantly higher than those who had undeclared infection (p = 0.024, 0.001 and 0.009, respectively) (Table 5). At age 38, 19.6% of women and 11.9% of men who did not report an infection were seropositive (Tables 4 and 5). At age 38, among seropositive individuals 63.3% (95% CI 54.4–71.4%) of women and 83.1% (95% CI 71.5–90.5%) of men gave no report of a Ct diagnosis (Tables 4 and 5).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 4. The Relationship between Ct Pgp3 antibody positivity and self-reported Ct, number of sexual partners, number of sexual partners of those who had not reported Ct infection and age of first coitus by age 26, 32 and 38 years. Women.

https://doi.org/10.1371/journal.pone.0151497.t004

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 5. The Relationship between Ct Pgp3 antibody positivity and self-reported Ct, number of sexual partners, number of sexual partners of those who had not reported Ct infection and age of first coitus by age 26, 32 and 38 years. Men.

https://doi.org/10.1371/journal.pone.0151497.t005

The proportion of seropositive women and men increased significantly with lifetime number of sexual partners at all three ages (p<0.001) (Tables 4 and 5). Limiting this analysis to those who never reported a Ct diagnosis, again a significant trend (p = 0.001) in seropositivity was found at all ages. At all ages, seropositivity was associated with younger first coitus in both women and men (p<0.05). This association was stronger in women than in men and did not vary by age, while in men the association weakened as they aged.

Pgp3 antibody persisted in most women and men over the 12-year study period (Fig 1). Of the 86 women who did not report a subsequent diagnosis of Ct (suspected re-infection) retested 12 years after their first positive test at age 26, 83 (96.5% [95% CI 90.1–99.3%]) remained seropositive. This included two who were weakly positive on ELISA at age 26 and were negative at age 32. Of the 37 men retested after 12 years, 31 (83.8% [95% CI 68.0–93.8%]) remained seropositive. There was no significant difference between the percentage of men and women remaining seropositive after 12 years (p = 0.12).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Flowchart showing the history of detectable Pgp3 antibody in women and men at age 26, 32 and 38 years.

The number of women and men testing Pgp3 antibody positive at age 26 is given. Shown at age 32 and 38 are the numbers of these positive individuals who: i) maintained Pgp3 seropositivity with no self-report of re-exposure to Ct, ii) remained Pgp3 positive, but also reported a re-infection with Ct, iii) had become seronegative, and iv) were missing at that time point. Finally, the numbers of women and men who became Pgp3 seropositive between age 26 and 32 together with their subsequent antibody status and infection history by age 38 are shown.

https://doi.org/10.1371/journal.pone.0151497.g001

Discussion

In the epidemiological cohort, Pgp3 double-antigen ELISA positivity was associated with a history of self-reported Ct diagnosis, with the proportion seropositive higher in women self-reporting Ct than men. Seropositivity increased with increasing number of sexual partners among those with and without a history of Ct, the latter suggesting undiagnosed Ct and/or unreported Ct diagnoses. Of the seropositive women at age 26 years, 96.5% still had detectable antibody after 12 years. While the numbers were relatively small, the results suggest more persistence of antibody among women than men, and no major decline in seropositivity over this period. Similar sensitivity was observed in women attending GUM clinics and in the Dunedin cohort at age 26, but not in men making estimates of cumulative risk of Ct infection using Pgp3 serology unreliable in men.

Well-characterised sera were used to determine assay sensitivity and specificity, enabling direct comparison with the previously developed indirect Pgp3 ELISA and commercial MOMP assays. We applied the assay to a cohort from which serial sera from the same individuals were tested blind. Sexual behaviour and self-reported STIs (including Ct) were collected at each assessment [19], reducing the risk of recall bias. Serum specimens were collected and stored without incident at -80°C, so it is unlikely that antibodies would have decayed over time. The cohort population tested in this study was broadly representative of the population in New Zealand at the time of its formation, and had a very high retention at ages 26, 32 and 38 years [30]. A weakness of this sample is that the number of people with a history of infection is small, resulting in relatively wide confidence intervals around positivity rates. In addition, Ct infection in the cohort was based on self-reports which under-estimate the number of actual infections, as many will be asymptomatic and unlikely to seek testing [31, 32]. Alternatively, some who report Ct infection may not have been infected.

Although the double-antigen ELISA demonstrated higher sensitivity than the indirect ELISA, it requires a 25-fold higher volume of serum. In the initial analysis of Ct-exposed and paediatric samples, no samples with indirect assay absorbance values <0.1 were positive when assayed by the double-antigen ELISA, nor did any with values >1.0 test negative. Further analyses of the Dunedin samples by the indirect ELISA (Table D in S1 Text) indicate that in large-scale studies the indirect ELISA is suitable for initial screening, and subsequent testing (sera with absorbance values between 0.1 and 1.0) by the double-antigen ELISA.

Public Health intervention strategies against Ct have been implicated in the rising number of infections on the basis that they undermine the development of protective immunity [33, 34]. Our findings lead us to the conclusion that, if arrested immunity occurs following early treatment, it does not substantially affect Pgp3 antibody development and persistence in women [6, 33, 34]. At age 26 years we observed in women a similar high detection of Pgp3 antibody (~80%) in Ct detection-positive GUM attendees, who are more likely to be incident cases [35], and the Dunedin cohort who are possibly more likely to be prevalent cases. Thus Pgp3 antibody could be used to estimate cumulative risk of infection in female populations in which control programmes are taking place [6, 36]. In an accompanying paper, Woodhall et al. use data and stored sera from nationally-representative household surveys from 1994 to 2012 in England to evaluate, using Pgp3 antibody the impact of widespread opportunistic Ct screening [36].

The higher proportion of women who are Pgp3 seropositive also holds true for MOMP peptide serology [9, 21, 37]. The mechanism for this is unclear, and merits further investigation [9, 21]. We observed a lower Pgp3 sensitivity, 25%, in men in the Dunedin cohort than the 54.4% observed in Ct detection-positive men attending GUM clinics. We believe the most likely explanation is poor diagnostic test accuracy in the 1990s resulting in some men in the Dunedin cohort being mis-diagnosed with Ct and/or recall bias [38]. If we consider only asymptomatic men with no history of infection attending GUM clinics, probably many with incident infection [35], the observed 25% sensitivity remains lower than the 43% observed in this GUM population. This low sensitivity cannot be explained by the hypothesis that early Ct treatment stops the development of Pgp3 antibody, as this would be more likely to affect the asymptomatic men attending GUM clinics than those in the Dunedin cohort [33, 35].

The clear increase of Pgp3 positivity with increasing number of sexual partners, a major risk factor for Ct and other STIs [39, 40], shows Pgp3 antibody can be used as a marker of Ct risk, especially among women. That this correlation persisted when restricted to those who did not report Ct is strong evidence for under-diagnosis and/or reporting in epidemiological studies, although not all seropositive results will be due to Ct infection as the specificity of the test is around 98%.

We observed that 63.3% women and 83.1% men who were seropositive at age 38 years did not report having ever been diagnosed with Ct, while around 2% may be false positives, this indicates that the majority of Ct infection was undetected/unreported in this cohort. Furthermore, there was an association of seropositivity with younger age of first sexual intercourse. This probably reflects a greater cumulative risk of Ct infection [41], but could also be accentuated by the possible increased biological susceptibility to Ct infection in younger women [42].

A key feature of this study was the repeated serological testing over a 12-year period. Among women reporting any past infection, 79.5% at age 26, 75% at age 32 and 74.6% at age 38 were antibody positive. While we have no information on when individuals became infected and could not examine the proportion of those who had seroconverted at some point in the past with subsequent seroreversion [7], we demonstrate antibody persistence over a 12-year period of 96.5% for women and 83.8% for men. While we excluded those reporting a subsequent Ct infection, we could not rule out unrecognised and/or reported reinfection that could explain the two women who were initially seropositive, negative six years later, then again positive 12 years later. The finding of persistence is consistent with previous analyses in which Pgp3 positivity in women was high soon after infection and remained stable at around 64% after a year [7]. While the point estimates suggest lower persistence for men, our numbers precluded examining this usefully. We are currently quantifying the loss of antibody since time of treatment in women attending departments of sexual health. This information will be used to develop the methodology in order to obtain estimates of Ct incidence and prevalence in England using serial population-based serum collections.

Conclusion

We show Ct infection was both common and usually undetected in women in this New Zealand cohort born in 1972 and 1973, with almost a third of women by this age having been infected by age 38, and demonstrate a strong and consistent correlation of Pgp3 seropositivity with self-reported Ct, multiple sexual partners and age of first intercourse. We conclude that our simple indirect assay facilitates high through-put antibody screening, and equivocal specimens can then be reconciled by the double-antigen ELISA.

Our findings suggest that surveys could make use of Pgp3 antibody in different age groups to examine the temporal change in cumulative risk of infection particularly among women, although persistence beyond 12 years has not yet been investigated [36]. These data provide a powerful argument for a role for Ct Pgp3 serology in evaluating Ct control programmes [36].

Supporting Information

Acknowledgments

We wish to thank the DMHDS members and their families for their long-term involvement, the current study director, Professor Richie Poulton, past director Dr Phil Silva, and all staff involved in data collection and other aspects of the study.