|Home | About | Journals | Submit | Contact Us | Français|
Cofactors might affect the risk of the rare progression from infection with carcinogenic human papillomavirus (HPV) to cervical premalignancy to invasive cancer. Some studies have observed that Chlamydia trachomatis infection is associated with increased risk for cervical cancer. In a large prospective cohort, we assessed the role of C trachomatis in cervical premalignancy and addressed confounding by HPV.
We identified 182 women with prevalent and 132 women with incident histological cervical intraepithelial neoplasia grade 2 (CIN2), grade 3 (CIN3), or cervical cancer (CIN2+) in the Costa Rica HPV Natural History Study. Control subjects were 995 (approximately 10% of the 10049) subjects who were randomly selected from the same study. Cervical HPV status at enrollment was determined by MY09/MY11 polymerase chain reaction amplification and dot-blot hybridization. The presence of C trachomatis DNA in cervical exfoliated cells at enrollment was determined by a novel serovar-specific polymerase chain reaction–based C trachomatis detection and genotyping assay. Plasma drawn at enrollment from each subject was used to determine C trachomatis immunoglobulin G (IgG) status. Logistic regression was used to examine the association between C trachomatis and CIN2+, taking into account possible confounding by HPV.
C trachomatis positivity at enrollment was associated with CIN2+ and concurrent and subsequent carcinogenic HPV infection. To account for confounding by HPV status, we restricted the analysis to women positive for carcinogenic HPV DNA at enrollment and found no association between C trachomatis status (as assessed by DNA or IgG) at enrollment and combined prevalent and/or incident CIN2+ (for C trachomatis DNA positivity, odds ratio = 0.77, 95% confidence interval = 0.42 to 1.41; for C trachomatis seropositivity, odds ratio = 1.09, 95% confidence interval = 0.85 to 1.41).
We found no association between C trachomatis status, as assessed by DNA or IgG, and risk of cervical premalignancy, after controlling for carcinogenic HPV-positive status. Previous positive associations between C trachomatis and cervical premalignancy could have been caused, in part, by an increased susceptibility to HPV infection.
Carcinogenic human papillomaviruses (HPVs) are causative agents for cervical cancer. It has been reported that Chlamydia trachomatis infection is associated with increased risk for cervical premalignancy.
Case–control study. Case subjects were women with prevalent or incident cervical intraepithelial neoplasia grade 2 or grade 3 or cervical cancer (CIN2+) from the Costa Rica HPV Natural History Study, and control subjects were also from this study. Cervical HPV status and C trachomatis status at enrollment was determined. The association between C trachomatis and CIN2+ was investigated, taking into account possible confounding by HPV.
Among all women in the study, C trachomatis positivity at enrollment was associated with CIN2+ and concurrent and subsequent carcinogenic HPV infection. When the analysis was restricted to women who were positive for carcinogenic HPV at enrollment to control for confounding by HPV status, no association was observed between C trachomatis status at enrollment and combined prevalent and/or incident CIN2+.
Previous finding between C trachomatis infection and cervical cancer could have been caused, in part, by an increased susceptibility to HPV infection.
No treatment data were available for C trachomatis or other sexually transmitted infections.
From the Editors
Although infection with carcinogenic human papillomavirus (HPV) is a necessary cause of cervical cancer (1,2), HPV infections are extremely common relative to the incidence of cancer (3). Cofactors might increase the risk of HPV-infected cells progressing to premalignancy and invasive cancer. Many studies (4,5,6,7,8,9), but not all of them (10,11), have observed that Chlamydia trachomatis is associated with cervical cancer or with persistent carcinogenic HPV types.
Although these results may reflect a causal association, the observed positive association between C trachomatis and cervical premalignancy and/or invasive cancer may be caused by residual confounding of other factors that are related to an HPV-positive status. Both C trachomatis and HPV are common sexually transmitted infections, and factors that are associated with acquisition, such as younger age and higher numbers of sexual partners, are shared by both. This strong relationship between HPV and C trachomatis can result in inadequate adjustment for HPV, especially when relying on HPV serology, which has low sensitivity, because many infected women do not seroconvert or revert to seronegativity.
To address the role of C trachomatis as a cofactor in cervical premalignancy and invasive cancer, we conducted a nested study of 314 case subjects of incident or prevalent cervical premalignancy and/or invasive cancer and an age-stratified random sample of 995 control subjects from the Costa Rica HPV Natural History Study. We measured C trachomatis infection with an assay for C trachomatis DNA in cervical tissue and with an assay for serum antibodies against C trachomatis. We accounted for HPV status by restricting the sample to carcinogenic HPV-positive case subjects and control subjects at enrollment.
The data to address the aims of this analysis are from the Costa Rica HPV Natural History Study, which is sponsored by the National Cancer Institute and described in detail elsewhere (12,13). Briefly, the primary aim of this population-based cohort study was to investigate the natural history of HPV infection and cervical intraepithelial neoplasia (CIN). Between June 22, 1993, and December 12, 1994, 10049 women who were 18 years or older, from random census sample of women from Guanacaste province in Costa Rica, were enrolled in the study, after providing informed consent. Participation among eligible women was more than 93%. Women were screened for cervical disease by use of multiple screening techniques and followed for 7 years, and those with evidence of possible high-grade disease were referred to colposcopy for further evaluation and treatment if indicated at any point during follow-up.
We excluded 92 virgin participants from the analysis (two were positive for C trachomatis by DNA or serology assay). For this analysis, 314 women with CIN2, CIN3, or invasive cervical cancer (CIN2+) histology (including 126 with CIN2, 138 with CIN3, and 50 with invasive cervical cancer) were considered to be case subjects. Cervical cancer in 22 of the 314 case subjects was detected by screening in the Natural History Study. An additional 28 were supplemental case subjects from the Guanacaste province who were diagnosed with cervical cancer during the same enrollment period in which the Natural History Study was conducted. These case subjects were initially identified from the National Tumor Registry and National Cytology Laboratory of Costa Rica, followed by review of their medical records to confirm their cancer status (14).
There were 182 case subjects with prevalent (diagnosed at baseline) and 132 case subjects with incident (diagnosed during follow-up) CIN2+. Control subjects in this analysis were an age-stratified random sample of 995 women (approximately 10%) originally selected from the 10 049 women in the Guanacaste baseline cohort for studies of other infections such as herpes simplex virus 2 and human herpesvirus 8 (15,16). We included only control subjects who did not develop a lesion during follow-up.
All study protocols were approved by a Costa Rican and the National Cancer Institute of the United States institutional review boards. All participants signed an informed consent form.
Trained interviewers administered a questionnaire on sociodemographic factors, sexual and reproductive history, contraceptive use, and smoking behavior at each visit. Blood (15 mL) was collected at each visit. In the field, samples were kept at 4°C in coolers, transported to the field station daily, and processed accordingly, and the plasma or serum, buffy-coat cells, and red blood cells were stored in separate 2-mL vials. These vials were frozen at −30°C at the field station and transported weekly to San Jose and stored at −70°C (13).
In addition, at each visit, exfoliated cells were collected from sexually active women for conventional and liquid-based cytology (ThinPrep; Cytyc Corporation, Marlborough, MA) with a Cervex brush (Unimar, Wilton, CT). For HPV typing, exfoliated cells were collected with a Dacron swab; these samples were stored in ViraPap DNA transport medium (Digene Corporation, now Qiagen, Gaithresburg, MD). Later in the study, the transport medium was changed to Digene’s DNA specimen transport medium. After collection, the cells in transport medium were kept in coolers at 4°C in the field. They were frozen at −30°C later that day and moved weekly to a −70°C freezer. Periodically, all samples (blood and cervical samples) were shipped on dry ice to the National Cancer Institute biorepository in Maryland. At this facility, samples were kept in a −70°C mechanical or liquid nitrogen (vapor phase) freezer until tested several years later (17). The 28 enrolled supplemental case subjects also received a pelvic examination when they enrolled in the study. At that time, cervical cells were collected and tested for HPV and other sexually transmitted infections as described above.
Cervical HPV DNA status was determined on enrollment exfoliated cervical tissue samples stored in specimen transport medium. All PCR testing and HPV genotyping were performed at the Albert Einstein College of Medicine, in New York City. An enhanced MY09/M11 L1 degenerate primer system with AmpliTaq Gold polymerase (TaqGold; Perkin-Elmer-Cetus, Norwalk, CT) was used to amplify HPV DNA, as previously described (17,18,19). The primer for MY09 was 5′-CGTCCMARRGGAWACTGATC-3′ and the primer for MY11 was 5′-GCMCAGGGWCATAAYAATGG-3′. Inclusion of human β-globin primers GH20 (forward primer sequence 5′-GAAGAGCCAAGGACAGGTAC-3′) and PC04 (reverse primer sequence 5′-CAACTTCATCCACGTTCACC-3′) amplified a cellular gene as a control for adequate and amplifiable sample DNA. Thermocycling conditions included initial denaturation at 95°C for 10 minutes; thereafter, each cycle consisted of 95°C for 60 seconds, 55°C annealing for 60 seconds, and extension at 72°C for 60 seconds, for a total 40 cycles with a final extension at 72°C for 5 minutes. After amplification, PCR products were analyzed for HPV DNA by electrophoresis and hybridization with radiolabeled generic HPV DNA probes. Type-specific oligonucleotide dot-blot hybridization was used for HPV type discrimination of the PCR products (oligonucleotide sequences are presented in Supplementary Table 1, available online). These products were tested for the following HPV genotypes: 2, 6, 11, 13, 16, 18, 26, 31–35, 39, 40, 42–45, 51–59, 61, 62, 64, 66–74, 81–85, and 89. The results of HPV typing were adjudicated by investigators who were masked to clinical outcomes. To confirm the sensitivity of the main HPV assay, approximately 2000 initially negative specimens, including those with and without cytological abnormalities, were retested with additional PCR primers (ie, GP5+/GP6+ and FAP). Few additional positive results arose from this confirmatory retesting.
HPV16 serological status was determined from plasma collected at enrollment by use of an enzyme-linked immunosorbent assay (ELISA) for HPV16 virus–like particles, performed at the Johns Hopkins Medical Institution (Baltimore, MD), as described previously (20,21,22). Briefly, ELISA microtiter plates were coated with HPV16 virus–like particles at 0.4–0.5 μg/mL of phosphate-buffered saline. The virus-like particles had been generated in insect cells from recombinant baculoviruses expressing the L1 genes of HPV16. After incubation and washing steps, plates were incubated further to block nonspecific antibody binding; the buffer was 0.5% (wt/vol) polyvinyl alcohol, with a molecular weight of between 30000 and 70000 (Sigma, St Louis, MO), in phosphate-buffered saline. Next, they were washed and incubated with plasma from participants, positive and negative quality control samples, and standards that were diluted at 1:100. After the washing steps, a horseradish peroxidase–conjugated anti-human immunoglobulin G (IgG) (Zymed, San Francisco, CA) was added to bind to the specific antibody. Color was developed by adding 2,2′-azino-di-(3-ethylbenzthiazoline-6-sulfonate) hydrogen peroxide solution (Kirkegaard and Perry, Gaithersburg, MD). Reactions were stopped, and optical density was read at 405 nm and at a reference wavelength of 490 nm. All specimens were tested in replicate on separate plates. Each batch included 2000–3000 specimens, along with appropriate interbatch and intrabatch reliability repeat specimens and specimens from a random sample of 200 of the 573 virgin participants in the study population. Results from the specimens from the virgin participants were used to calculate the ELISA cutoff. The cutoff for the HPV16 virus–like particle ELISA serology was calculated by comparing the distribution of the concurrently tested specimens from the virgin participants in that batch, independently for each test batch. Seropositivity was defined as five standard deviations greater than the mean obtained for the concurrently tested virgin participants. Because aliquots of the same samples were tested twice in this analysis, a stringent definition of HPV16 seropostivity was used that was based on the results from both replicates (ie, positive only if they were highly positive in either of the two and not negative in either; negative only if they were negative in both experiments or negative in one experiments and low positive in the other; and equivocal if they were positive on one experiments but negative or equivocal on the other).
C trachomatis DNA was detected and genotyped in enrollment cervical samples that were collected in 1993 and 1994 by use of the C trachomatis amplification, detection, and genotyping assay (Labo Biomedical Products BV, Rijswijk, the Netherlands) (23,24). Briefly, total DNA was isolated from a 200-μL PreservCyt aliquot of a cervical sample with the MagNA Pure LC instrument and the Total DNA isolation kit (Roche Diagnostics, Almere, the Netherlands). After isolation, the DNA was eluted from the magnetic beads and resuspended in 100 μL of elution buffer (Roche Diagnostics). Exact concentrations differ in each sample because the sample in each cervical swab was taken from an inhomogeneous matrix. Because the PCR target is a C trachomatis DNA, the exact number of these DNA molecules cannot be determined.
Amplification, detection, and genotyping steps were performed according to instructions from the manufacturer of C trachomatis detection and genotyping assay (Labo Biomedical Products BV), as described previously (23,24). Briefly, a 10-μL aliquot of isolated DNA was amplified by PCR targeting both genomic and cryptic plasmid C trachomatis DNA. This multiplex PCR of nine primers amplifies a fragment of 241 base pairs from the cryptic plasmid and a fragment of 160 or 157 base pairs from variable region 2 of the Omp1 gene (Supplemental Table 2, available online). The standard PCR program involves a 9-minute preheating step at 94°C, followed by 40 cycles of amplification (30 seconds at 94°C, 45 seconds at 55°C, and 45 seconds at 72°C) and a final 5-minute elongation at 72°C.
After the PCR, C trachomatis DNA was detected by C trachomatis–specific probe hybridization in a DNA enzyme immunoassay that used 10 μL of the PCR product. The mixture of probes present in the C trachomatis DNA enzyme immunoassay can recognize all C trachomatis serovars and genovariants that have been deposited in GenBank. Briefly, reverse primers contained a biotin label at the 5′ end, which can bind to streptavidin-coated wells of a microtiter plate. Captured amplimers were denatured in denaturation solution (alkaline solution (NaOH) containing EDTA provided in the kit. Uncaptured DNA strands were removed by washing and then a defined mixture of digoxigenin-labeled probes, which hybridize to the captured strands of the PCR products, was added. The hybrids were detected by an enzymatic color reaction, as measured at an optimal density of 450 nm. The optical density value of each well was compared with that of a titrated positive sample, a borderline positive sample, and a negative control sample, all of which were in the kit. All C trachomatis DNA-positive samples were further genotyped with a C trachomatis DNA genotyping reverse hybridization assay, containing a probe for the cryptic plasmid, probes for the three C trachomatis serogroups (B, C, and I), and probes for the 14 serovars (A, B, or Ba; C, D, or Da; E, F, G, or Ga; H, I, or Ia; J, K, L1, L2, or L2a; and L3). The genotyping step was performed according to the manufacturer’s instruction. Briefly, 10 μL of PCR product was mixed with 10 μL of denaturation solution in a plastic trough containing the strips for 5 minutes at room temperature; further incubations and washing steps were performed automatically in an AutoLiPA instrument (Tecan Austria GmbH, Salzburg, Austria). After hybridization, washing, and incubating with alkaline phosphatase–streptavidin conjugate, the amplimer-probe hybrids were visualized by use of a substrate buffer that generated a purple line at the location of a specific probe. In general, three probe reactions occur per C trachomatis serovar (ie, reaction with the cryptic plasmid probe, a serogroup probe, and a serovar probe). These lines were visually interpreted.
Serum samples collected at enrollment were tested for C trachomatis antibody with a commercially available C trachomatis IgG and IgA pELISA assay (Medac Diagnostika, Hamburg, Germany). IgA antibodies against C trachomatis in serum were also determined to ensure detection of early-stage C trachomatis infections that have not yet developed an IgG response. The tests were performed according to the manufacturer’s instructions. Briefly, serum samples were diluted 1:50 with sample diluents (phosphate-buffered saline, Tween 20, and newborn calf serum, all provided as part of the commercial kit, at pH 7.0–7.2), and 50 μL of diluted serum was added per well of a 96-well microplate and incubated for 60 minutes at 37°C. Plates were washed three times with 200 μL of wash buffer (10× phosphate-buffered saline and Tween 20 at pH 7.2–7.4, provided as part of the kit), 50 μL of conjugate (containing goat anti-IgG) was added, and the plates were incubated for 60 minutes at 37°C and washed three times with wash buffer. Finally, 50 μL of the tetramethylbenzidine solution from the kit, the substrate for the color reaction, was added and incubated for 30 minutes at 37°C. The reaction was stopped by adding 100 μL of stop solution (0.5 M sulfuric acid) to each well, and the optical density at 450 nm was read for each well. The quantitative cutoff index was defined as follows: negative = less than 0.9; equivocal = 0.9–1.1; positive = greater than 1.1. The 19 samples with equivocal IgG results were excluded from the analyses.
We examined associations between the different methods for determining C trachomatis status (ie, C trachomatis DNA as measured by DNA enzyme immunoassay and serological determination of IgG and IgA against C trachomatis; Table 1). Agreement between C trachomatis DNA results and either serological measure (IgG kappa = 0.19 and IgA kappa = 0.09) was low. Agreement between serological results for IgG and IgA against C trachomatis was also low (kappa = 0.27). Overall, there were only 74 anti-C trachomatis IgA-positive samples, of which 77% (57 of the 74 samples) were also IgG positive. However, in an ancillary analysis, we considered positivity to be a composite variable defined as a positive result for any of the C trachomatis serological markers or for C trachomatis DNA and found that the results were not altered by use of the composite variable. Hence, for the remainder of the analyses, we have presented only the C trachomatis IgG data and the C trachomatis DNA data.
To examine the extent of confounding, we investigated the enrollment characteristics and their association with C trachomatis infection (separately for DNA and serological positivity) among control subjects by use of univariate and multivariable logistic regression models to estimate odds ratios (ORs), adjusted odds ratios (aORs), and 95% confidence intervals (CIs). Variables were derived from medical history forms and questionnaires. Although we examined many potential associations, the following are not presented in the tables because they did not affect the strength of the associations investigated: education; age at menarche; ever pregnant; age at first pregnancy; total number of pregnancies, stillbirths, abortions, and/or miscarriages; ever use of an intrauterine device; tubal ligation; and PAP cytology.
Finally, to examine the association between C trachomatis and CIN2+, taking into account possible confounding by HPV, we used multivariable logistic regression among carcinogenic HPV-positive women only (defined as women with HPV types 16, 18, 31, 33, 35, 39, 45, 51, 52, 56, 58, 59, 66, and/or 68). In ancillary analyses, we considered the association between genotype-specific C trachomatis and CIN2+. Additionally, we examined the effect stratifying the analysis by other factors, including incident and prevalent CIN2+, excluding CIN2+ from the outcome definition and limiting the analysis to CIN3+ case subjects, and considering possible effect modification by age (<30 years or ≥30 years). We also evaluated the role of enrollment C trachomatis on subsequent new detection of carcinogenic HPV. The outcome of interest was defined as any cervical carcinogenic HPV infection newly detected (referred to as incident infection) during follow-up. The main exposure of interest was enrollment C trachomatis as measured by DNA and serology. Thus, we compared the cumulative incidence rate of incident carcinogenic HPV infection among C trachomatis–negative participants and among C trachomatis–positive participants (with results for both DNA and IgG presented). Incidence rates and rate ratios (RRs) of newly detected carcinogenic HPV infection were calculated with person-time methods for respective C trachomatis–positive women relative to C trachomatis–negative women (separately by DNA and IgG results).
All statistical analyses were performed with SAS software, version 9.1 (SAS Institute, Cary, NC). All statistical tests were two-sided.
C trachomatis DNA test results were available for 1230 (94%) of the 1309 subjects. The main reason for missing results (in 30 case subjects and 49 control subjects) was lack of an adequate sample for testing. C trachomatis serology results were available for 1283 (98%) of the 1309 subjects.
Selected enrollment characteristics among case subjects and control subjects are shown in Table 2. Case subjects were statistically significantly younger than control subjects; the median age of case subjects was 35 years (interquartile range = 27–46 years) and that of control subjects was 44 years (interquartile range = 31–56 years) (P < .001). Case subjects were statistically significantly more likely than control subjects to have initiated sex at an earlier age (P = .02), had more lifetime (P < .001), and recent (P = .01) sex partners and were more likely to be infected with C trachomatis (P = .004 for DNA and P = .02 for IgG). Thus, as expected, case subjects had a higher risk profile for sexually transmitted infections as judged by demographic and sexual factors than control subjects.
Demographic and behavioral characteristics at enrollment and their association with C trachomatis status (DNA and serology separately) among control subjects are presented in Table 3. A positive C trachomatis DNA status was statistically significantly associated with younger age and decreased with increasing age (aOR = 0.97, 95% CI = 0.95 to 0.99). A positive C trachomatis DNA status was statistically significantly associated with having two or three sex partners in the past year compared with no partners in the past year (aOR = 5.43, 95% CI = 1.03 to 28.72) and with a positive carcinogenic HPV DNA status (aOR = 2.47, 95% CI = 1.34 to 4.56). C trachomatis serological status was not associated with age; however, women with a positive C trachomatis serological status, compared with those with a negative status, were more likely to have had an ectopic pregnancy (aOR = 5.88, 95% CI = 1.13 to 30.53), smoked cigarettes (aOR = 2.16, 95% CI = 1.34 to 3.48), had positive serology for herpes simplex virus 2 (aOR = 1.90, 95% CI = 1.30 to 2.79), and to be seropositive for HPV16 (aOR = 2.03, 95% CI = 1.40 to 2.94). Thus, a positive C trachomatis DNA status was associated with an active C trachomatis infection, and a positive C trachomatis serological status was reflective of a past exposure to C trachomatis.
Relationships between C trachomatis status, as assessed by DNA or serological assays, and CIN2+ among all subjects at enrollment and among those who were either positive or negative for carcinogenic HPV DNA at enrollment are shown in Table 4. When HPV status was not controlled and case subjects were compared with control subjects, C trachomatis status, as measured by DNA or serology, was statistically significantly associated with CIN2+ (for C trachomatis DNA, OR = 1.78, 95% CI = 1.20 to 2.65; for C trachomatis IgG, OR = 1.24, 95% CI = 1.07 to 1.44). We next stratified the analysis by enrollment carcinogenic HPV status. Among carcinogenic HPV-negative subjects, we found a strong and statistically significant association between a positive C trachomatis DNA status at enrollment and incident CIN2+ (OR = 2.19, 95% CI = 1.03 to 4.66) but no association between a positive C trachomatis IgG status and CIN2+. It should, however, be noted that there were few carcinogenic HPV-negative subjects with CIN2+.
To account for confounding by HPV, we restricted the analysis to subjects whose exfoliated cervical cells at enrollment were positive for carcinogenic HPV, as determined by a positive HPV DNA test result (Table 4). Among subjects in this group, no association was found between CIN2+ and C trachomatis (for positive C trachomatis DNA status, OR for CIN2+ = 0.77, 95% CI = 0.42 to 1.41; for positive C trachomatis IgG status, OR for CIN2+ = 1.09, 95% CI = 0.85 to 1.41). We also investigated this relationship among women who were positive for HPV16 DNA and/or for HPV16 IgG but found no association between CIN2+ and C trachomatis (data not shown).
A positive C trachomatis status at enrollment was also associated with incident carcinogenic HPV infection (RR for incident carcinogenic HPV = 2.6, 95% CI = 1.6 to 4.2, when subjects with a positive C trachomatis DNA status were compared with those with a negative C trachomatis DNA status; RR = 1.6, 95% CI = 1.1 to 2.4, when subjects with a positive C trachomatis IgG status were compared with those with a negative C trachomatis IgG status) (Table 5).
When we restricted the analysis to the 188 case subjects with CIN3 or cancer (a more stringent case definition) and the control subjects to the 995 control subjects only or the control subjects plus the CIN2 cases (n = 1121), no association was found between C trachomatis status and CIN3+ among HPV-infected women. Finally, when we restricted the analysis to the 50 case subjects with cancer and the 995 control subjects (the random sample of control subjects from the baseline population), no association was found between C trachomatis status and cancer among HPV-infected women (data not shown).
We further investigated the relationship between C trachomatis and CIN2+ among carcinogenic HPV-positive women who were stratified by age (<30 years or ≥30 years). No association was found among women younger than 30 years between C trachomatis DNA status and CIN2+ (OR = 1.13, 95% CI = 0.45 to 2.77) or between C trachomatis IgG status and CIN2+ (OR = 1.28, 95% CI = 0.83 to 1.99). In addition, no association was found among women aged 30 years or older between C trachomatis DNA status and CIN2+ (OR = 0.61, 95% CI = 0.27 to 1.39) or between C trachomatis IgG status and CIN2+ (OR = 1.00, 95% CI = 0.74 to 1.37).
We next investigated the association between CIN2+ and C trachomatis serogroups, results of which are shown in Table 6. Among subgroups of subjects who were positive for a specific serogroup of C trachomatis, no statistically significant association was found between a positive status for a C trachomatis serogroup and CIN2+. Although the number of subjects in each subgroup were small, we observed a non-statistically significant association between a positive status for C trachomatis serogroup I and an increased risk for CIN2+ (OR = 3.26, 95% CI = 0.37 to 28.85) and a statistically significant protective association between a positive status for C trachomatis serogroup B and a decreased risk for CIN2+ (OR = 0.33, 95% CI = 0.14 to 0.80).
In this study from the large population-based Costa Rica HPV Natural History Study, we did not find an association between C trachomatis infection and CIN2+ among subjects with carcinogenic HPV DNA at baseline. Determining the role of C trachomatis as a cofactor for cervical premalignancy and/or invasive cancer requires that HPV status and the natural history of cervical cancer be taken into consideration. The multistage development of invasive cervical cancer from HPV infection takes many years. In general, HPV infections occur close to the age of sexual debut, more than 90% disappear within 2–5 years, and most women who are diagnosed with cervical cancer are aged 40 years or older (3,25).
HPV is the most prevalent viral sexually transmitted infection, C trachomatis is the most prevalent bacterial sexually transmitted infection, and co-infections with both are common (26). We showed that C trachomatis status at enrollment was a risk factor for current and subsequent carcinogenic HPV infection (RR = 2.6, 95% CI = 1.6 to 4.2). It is therefore possible that C trachomatis is associated with cervical cancer because it is associated with HPV acquisition (either by a noncausal link through common risk factors, such as infected partners and sexual behaviors, or by a causal disruption of the epithelial tissue). However, on the basis of our results, it appears to be unlikely that C trachomatis infection affects HPV persistence and progression to cervical premalignancy because we found no association between C trachomatis status and CIN2+ or between C trachomatis status and CIN3+ in our analysis after adjusting for HPV (by restricting the analysis to carcinogenic HPV-infected women).
Whether C trachomatis acts as a cofactor in the steps from progression from premalignancy to invasive cervical cancer cannot be adequately addressed by results of this study because the study was not sufficiently powered to evaluate the role of C trachomatis in invasive cervical cancer. The lack of association between C trachomatis and invasive cervical cancer observed in this analysis is contrary to findings of the pooled study of case–control studies of invasive cervical cancer from International Agency for Research on Cancer (IARC) (4). They found a positive association between C trachomatis serology and invasive cervical cancer among HPV-positive women. The IARC study was large and had a well-designed epidemiological and laboratory component. However, the lack of prospective information for control subjects in that study means that HPV status and C trachomatis status at relevant times were not available.
Our results also differ from other studies that found a positive association between C trachomatis status and invasive cervical cancer, after adjusting for HPV seropositivity in a statistical model (6,7,27–30) or using HPV serostatus as a marker of HPV exposure among control subjects (8). Retrospective control for HPV by use of HPV serology is insensitive and thus not an ideal adjustment in examining a possible link between C trachomatis and cervical cancer. Because HPV is a necessary etiologic agent for cervical cancer and thus a strong confounder, even weak associations with a positive HPV status can result in the identification of a spurious epidemiological association. Although HPV serology is thought to be a fairly specific surrogate for cumulative HPV exposure (31), it has low sensitivity because not all HPV-infected women seroconvert. Additionally, the HPV serology tests used in some previous studies targeted only a few HPV types (7). Hence, observed positive associations might have resulted from residual confounding caused by misclassification of HPV status.
Moreover, the observation that a positive C trachomatis status at enrollment was associated with an increased risk for subsequent carcinogenic HPV infection may explain why C trachomatis was also associated with an increased risk for CIN2+ among initially HPV-negative women. Adjustment for HPV status at enrollment eliminated the association between a positive C trachomatis status and CIN2+, perhaps because a positive C trachomatis status might be associated with an unknown or unmeasured set of behaviors or sexual partner characteristics that lead to both future and concurrent HPV infection. This consideration is important when evaluating retrospective studies that used HPV status at a single time point in their adjustment for HPV, especially among control subjects.
Two interesting observations in our study were the association between a positive C trachomatis status and an incident CIN2+ among subjects who initially had a negative HPV status and the lack of an association among subjects who initially had a positive HPV status. These results indicate that stratum-specific analyses provided greater insight into the relationship between C trachomatis and CIN2+. The assays that we used to determine C trachomatis status differed from those used in previous studies (4,5,6,7,29,32). Previous studies used the micro-immunofluorescent assay to determine C trachomatis serological status; although this assay is considered to be the gold standard, it is subjective and it requires considerable experience to differentiate between strains, and labor intensive particularly for large epidemiological studies. A strength of this study was that we used two different assays to measure C trachomatis status. 1) The serological assay that we used as a measure of past C trachomatis exposure was very specific for C trachomatis. 2) The C trachomatis DNA assay examined current C trachomatis infection at the cervix. This assay provided a novel aspect for this study. Both these assays have similar or higher sensitivity and specificity compared with other commercial assays (23,24). Although we mention these differences in the assays used for completeness of discussion, we do not believe that the differences were large enough to have resulted in the discrepant finding between studies. To ensure that our measures of C trachomatis were accurate, we examined determinants of C trachomatis DNA and seropositivity among the control subjects and showed that C trachomatis DNA status was independently associated with several sexual risk factors, and C trachomatis IgG status was independently associated with ectopic pregnancy.
We constructed a model by statistically adjusting for HPV16 serology (dichotomous negative vs. positive test results, excluding equivocal test results), sexual risk factors (including lifetime number of sexual partners on a continuous scale), and enrollment age (on a continuous scale) to investigate the association between C trachomatis (DNA and/or seropositivity, to try to increase our power), and CIN2+, to replicate previous studies (6,7,9,28,29). We observed an increased risk of C trachomatis infection among case subjects compared with control subjects (OR = 1.35, 95% CI = 1.01 to 1.82); however, when we restricted the analysis to HPV16-seropositive subjects (not including equivocals), we found no association (OR = 0.84, 95% CI = 0.48 to 1.48) (data not shown).
The number of subjects with specific C trachomatis DNA serovars was too small to examine the association between each serovar and CIN2+, and so we investigated the association at the C trachomatis serogroup level. One notable finding of this study was a non-statistically significantly increased association between the non-B serogroups and CIN2+, indicating that, if C trachomatis is a cofactor in CIN2+, then C trachomatis would belong to one of the non-B serogroups. These data are in concordance with the association of serovar G antibodies and squamous cell carcinoma of the cervix, as described previously (6). The most prevalent serovars among women in Costa Rica belong to the serogroup B (specifically, serovars E and D or Da) and the intermediate serogroups (serovar F) (23). Interestingly, we found a statistically significant protective association between serogroup B and CIN2+ among HPV-positive women. These findings were based on very few subjects and need to be validated in larger cohorts.
Our study was limited by lack of data on treatment for C trachomatis. Because this analysis was conducted after the completion of the cohort study, in which C trachomatis testing was not part of the protocol, women might have sought treatment for C trachomatis on their own. In addition, no information on treatment for other sexually transmitted infections was collected.
In summary, we found no association between C trachomatis status, as assessed by DNA or serology, and risk of cervical premalignancy after controlling for carcinogenic HPV-positive status. Thus, previous reports of an association between C trachomatis and cervical premalignancy may be, in part, caused by confounding by HPV status or by an increased susceptibility to HPV infection among women with a positive C trachomatis status.
National Institutes of Health (N01-CP-21081, N01-CP-33061, N01-CP-40542, N01-CP-50535, N01-CP-81023, and CA78527 to R.D.B.). The Guanacaste cohort (design and conduct of the study, sample collection, management, analysis, and interpretation of the data) for the enrollment and follow-up phases were supported by the National Cancer Institute, National Institutes of Health, Department of Health and Human Services.
M. Safaeian and K. Quint contributed equally to this work.
The authors take full responsibility for the design of the study; the collection, analysis, or interpretation of the data; the writing of the manuscript; and the decision to submit the manuscript for publication.