Anttila T, Saikku P, Koskela P, Bloigu A, Dillner J, Ikäheimo I, Jellum E, Lehtinen M, Lenner P, Hakulinen T, Närvänen A, Pukkala E, Thoresen S, Youngman L, Paavonen J. Serotypes of Chlamydia trachomatis and Risk for Development of Cervical Squamous Cell Carcinoma. JAMA. 2001;285(1):47–51. doi:10.1001/jama.285.1.47
Author Affiliations: National Public Health Institute, Oulu, Finland (Drs Anttila and Koskela and Ms Bloigu); Department of Medical Microbiology, University of Oulu, Oulu, Finland (Dr Saikku); Microbiology and Tumor Biology Center, Karolinska Institute, Stockholm, Sweden (Dr Dillner); Microbiology Laboratory, Oulu University Hospital, Oulu, Finland (Dr Ikäheimo); Institute of Clinical Biochemistry, Rikshospitalet, Oslo, Norway (Dr Jellum); School of Public Health, University of Tampere, Tampere, Finland (Dr Lehtinen); Department of Oncology, University of Umeå, Umeå, Sweden (Dr Lenner); Departments of Public Health (Dr Hakulinen) and Obstetrics and Gynecology (Dr Paavonen), University of Helsinki, Helsinki, Finland; Department of Biochemistry, University of Kuopio, Kuopio, Finland (Dr Närvänen); Finnish Cancer Registry, Helsinki (Dr Pukkala); The Cancer Registry of Norway, Oslo (Dr Thoresen); and Clinical Trial Service Unit, University of Oxford, Oxford, England (Dr Youngman).
Context Human papillomavirus (HPV) infection has been established as a cause
of cervical cancer. Epidemiologic studies suggest that Chlamydia trachomatis infection also confers increased risk for cervical
squamous cell carcinoma (SCC). Whether this risk is serotype-specific is unknown.
Objective To study the association between exposure to different C trachomatis serotypes and subsequent development of cervical SCC.
Design and Setting Longitudinal, nested case-control study within a cohort of 530 000
women who provided samples to serum banks in Finland, Norway, and Sweden.
The data files were linked to respective national cancer registries.
Subjects One hundred twenty-eight women who had developed invasive cervical SCC
at least 12 months following serum donation. Each case had 3 matched controls.
Main Outcome Measure Risk for the development of cervical SCC by IgG antibodies to 10 different C trachomatis serotypes, adjusted for antibodies to HPV
types 16, 18, and 33 and for serum cotinine levels.
Results Of specific C trachomatis serotypes, serotype
G was most strongly associated with SCC (adjusted odds ratio [OR], 6.6; 95%
confidence interval [CI], 1.6-27.0). Other serotypes associated with SCC were
I (OR, 3.8; 95% CI, 1.3-11.0) and D (OR, 2.7; 95% CI, 1.3-5.6). Presence of
serum IgG antibodies to more than 1 serotype increased the adjusted ORs for
SCC (P<.001 for trend).
Conclusions Chlamydia trachomatis serotype G is most strongly
associated with subsequent development of cervical SCC. Increasing numbers
of exposures to different C trachomatis serotypes
also increases risk. Our results strengthen the evidence that there is a link
between past C trachomatis infection and cervical
Human papillomavirus (HPV) infection is the leading cause of cervical
neoplasia.1- 3 Additional
risk factors include other sexually transmitted infections (STIs) and smoking.
The evidence linking oncogenic HPV types and cervical carcinoma is very strong
and consistent. Human papillomavirus DNA–based cohort studies4,5 have confirmed the seroepidemiologic
findings6,7 that past HPV infection
predisposes women to developing cervical carcinoma. Longitudinal seroepidemiologic
studies have also provided evidence that Chlamydia trachomatis infection is an independent risk factor for the development of invasive
cervical squamous cell carcinoma (SCC).6- 8
Cervical chlamydial infection can persist for long periods of time.9 Similarly, elevated antichlamydial antibody titers
persist for several years.10 Microimmunofluorescence
(MIF) testing is still the gold standard for chlamydia serology, and researchers
also can use MIF for serotyping.11,12
We found a link between the presence of serum antibodies to C trachomatis and the subsequent development of cervical SCC.8 This study was conducted to determine whether this
association is serotype-specific.
The serum banks and cancer registries we used have been described in
detail.7,8 Our study used a joint
cohort of 3 population-based serum banks to which a total of 530 000
women have donated blood samples. The Finnish Maternity Cohort13
has collected samples since 1983 from more than 98% of pregnant women in Finland.
The Janus project14 collected samples from
1973-1987 and from 1987-1991 during routine health or screening examinations
in various counties in Norway. The Västerbotten project15
has collected samples since 1986 from about 65% of adults during a health
promotion project in northern Sweden.
The Finnish Cancer Registry and the Cancer Registry of Norway are nationwide,
while the Regional Cancer Registry at the Oncological Centre in Umeå
covers the 4 northernmost counties of Sweden.16
These registries achieve almost 100% capture by using reports from hospitals,
pathology laboratories, and physicians.
Women with cervical carcinoma were identified by linking the data files
of the population-based serum banks with the nationwide cancer registries,
as described.7,8 By the end of
1994 and following histological review, 181 women with invasive cervical carcinoma
were identified. For each case, the earliest prediagnostic serum sample was
chosen and 3 matched controls free of cancer at the time of case diagnosis
were randomly selected. Controls were matched for sex, age at serum sampling
(±2 years), storage time of serum samples (±2 months), country,
and (in Norway) county. Serum sample volume was insufficient in 10 controls.
Thus, the final number of controls was 533. Permission to link the Finnish
Maternity Cohort serum bank and the cancer registry was obtained by the Data
Inspection Board of Finland, the ethics committees of the National Research
and Development Center for Health and Welfare (STAKES), Finland, and the University
Hospital, Helsinki, Finland. In Norway and Sweden, all serum bank donors provided
Serum IgG antibodies to C trachomatis and Chlamydia pneumoniae were measured by MIF testing as described.7,8 Titers of ≥16 were considered positive
for C trachomatis. Titers of ≥32 were considered
positive for C pneumoniae, which was used as a control
antigen. Serum IgG antibodies to C trachomatis were
also measured by 2 types of enzyme-linked immunosorbent assay (ELISA): an
elementary body (EB) ELISA (Chlamydia IgG, Labsystems Co, Helsinki, Finland),
and a peptide ELISA (Chlamydia trachomatis IgG, Labsystems Co). The latter
applies C trachomatis major outer membrane protein
(MOMP) variable domain IV synthetic peptide as the antigen.17
Microimmunofluorescence was used to further analyze all C trachomatis antibody-positive serum samples for antibodies against
the following C trachomatis serotypes: B, D, E, F,
G, and J (American Type Culture Collection, Rockville, Md); and C, H, I, and
K (Washington Research Foundation, Seattle, Wash). Elementary body antigens
were prepared from C trachomatis serotypes B, D,
E, F, G, and J (grown in McCoy cells), and from serotypes C, H, I, and K (grown
in HeLa-229 cells), and purified using conventional techniques.18
Elementary body aliquots were stored at −70°C prior to use. Type-specificity
of the different antigen preparations was confirmed by C trachomatis type-specific monoclonal antibodies. Immunoglobulin G
antibodies to single serotypes (B, D, E, F, G, J, C, H, I, and K) were measured
by MIF as described above. The serum samples were analyzed at 2-fold dilutions.
For this study, diagnostic-phase tumor biopsy specimens were available
from 85 (47%) of the 181 cases. Two to three 10-µm sections were used
for DNA extraction from paraffin-embedded biopsy specimens.19
For each block, the microtome was cleaned with alcohol and a new blade used.
As a contamination control, sections were taken from empty paraffin blocks
between each specimen. To remove the paraffin, the sections were rinsed with
xylene and centrifuged (2 times). To remove the xylene, the sections were
then rinsed with 96% alcohol, centrifuged, and decanted (3 times). Acetone
was then added and the sections incubated at 56°C for 1 hour, after which
100 µL of proteinase K (200 µL/mL) was added for an overnight
incubation at 56°C. The next day the tubes were centrifuged, decanted,
and incubated at 95°C for 15 minutes. Finally, the tubes were centrifuged
and the DNA concentration was measured by a spectrophotometer. Seventy-nine
(93%) of the 85 cervical biopsy specimens analyzed showed successful amplification
of the human DQA gene. The presence of C trachomatis
DNA was determined using the automated Cobas Amplicor Chlamydia
trachomatis test (Roche Molecular Diagnostics, Branchburg, NJ) from
all cases that showed successful amplification of the human HLA DQA gene.
The test uses primers CP24 and CP27 to define a DNA sequence of approximately
207 nucleotides within the cryptic plasmid of C trachomatis. An internal control was added to the Cobas Amplicor test to identify
processed specimens containing substances that interfere with amplification.
Immunoglobulin G antibodies to HPV types 16, 18, and 33 were determined
by ELISA using viruslike particle capsid antigens, as described in other studies.7,8 Human papillomavirus ELISA has been
extensively validated in previous studies.20,21
Serum cotinine was used as a surrogate marker for smoking: levels were measured
Odds ratios (ORs) with 95% confidence intervals (CIs) and 2-sided P values were estimated by conditional logistic regression
for matched case-control sets.22 The effect
of smoking and infection with HPV-16, HPV-18, or HPV-33 was considered by
adjusting for serum cotinine levels and serum antibodies to any of these HPV
types. Test for trend was calculated by modeling the number of serotypes as
1 quantitative variable, assuming scores 0 to 4 for exposure to 0, 1, 2, 3,
or 4 or more serotypes.
Of the 181 patients with invasive cervical carcinoma, 48 were from Finland,
129 from Norway, and 4 from Sweden. Of all carcinomas, 109 (60%) were localized,
and 62 (34%) were metastatic. In 10 cases (6%) the stage was unknown. The
mean age of the patients at diagnosis was 44 years (range, 23-64 years), and
the mean time between serum donation and diagnosis was 56 months (range, 1-221
months). Of all patients, 150 had cervical SCC, of whom 128 had a time between
serum donation and diagnosis (lag time) of at least 12 months (Table 1).
The overall prevalence rates of serum IgG antibodies to C trachomatis among all cases and controls were 27% and 13%, respectively
(Table 1). The corresponding case
and control rates for HPV-16, HPV-18, or HPV-33 were 37% and 18%, respectively,
and for cotinine, 50% and 39%. Antibodies to C trachomatis were analyzed by 3 methods: MIF and 2 commercially available ELISAs
(Table 1). In general, ELISAs
were more sensitive than MIF. Among all invasive carcinomas (n = 181), antibodies
measured by the ELISAs yielded low point estimates for the risk of cervical
carcinoma (EB ELISA: OR, 1.5, 95% CI, 1.0-2.4; peptide ELISA: OR, 1.3, 95%
CI, 0.9-1.9; both adjusted for HPV types 16, 18, and 33 and serum cotinine).
Antibodies measured by MIF were associated with higher risk (adjusted OR,
1.8; 95% CI, 1.1-2.8). Serum IgG antibodies to C pneumoniae were not associated with cervical carcinoma (adjusted OR, 1.2; 95%
The highest point estimates were found for SCC diagnosed 12 or more
months after serum donation. Within that group (n = 128), serotype G was most
strongly associated with SCC (adjusted OR, 6.6; 95% CI, 1.6-27.0) (Table 1). Other serotypes also associated
with SCC were I (OR, 3.8; 95% CI, 1.3-11.0) and D (OR, 2.7; 95% CI, 1.3-5.6)
(Table 1). In addition, there
was a borderline association with serotype B (OR, 4.1; 95% CI, 1.0-18.0).
Exposure to more than 1 serotype increased the risk for cervical SCC (P<.001 for trend) (Table 2).
Chlamydia trachomatis DNA was detected by polymerase
chain reaction (PCR) in 4 (5%) of the 79 cases analyzed. One of the DNA-positive
cases was antibody positive for multiple serotypes (D, E, F, H, I, J), whereas
the other cases were C trachomatis seronegative.
To the best of our knowledge, this is the first study providing longitudinal
seroepidemiologic evidence of an association between exposure to specific
serotypes of C trachomatis and cervical SCC. Presence
of serum IgG antibodies to C trachomatis serotype
G was associated with the highest risk. Immunoglobulin G antibodies to more
than 1 serotype of C trachomatis increased the risk
for subsequent development of SCC.
Microimmunofluorescence is the method of choice for C trachomatis serotyping.12 So far,
18 different serotypes (or serovars) have been described.11,23
Distribution of the genital serotypes varies from one area to another, suggesting
that some serotypes have biological advantage over others in defined populations.24 Serotypes D and E represent approximately 50% of
all isolates, followed by F and G serotypes, which represent 15% to 40%; other
serotypes represent less than 10% each.25- 27
Serotypes E and G have been found more often in women than men, whereas serotype
D has been found more frequently in men than in women.28,29
In another seroepidemiologic study from Finland, antibodies to the GFK
serotype pool were more common in women who developed SCC than in controls.6 Similarly, serotype G was the strongest risk factor
for SCC in the present study. Serotype G has also been associated with symptomatic
infections and upper genital tract infections.30,31
Serotype D was also associated with SCC in this study. When different serotypes
of C trachomatis were inoculated intravaginally into
mice, the duration of infection with D and E was longest and induced the highest
antibody titers.32 Thus, specific C trachomatis serotypes might be more virulent, perhaps less sensitive
to appropriate antimicrobial treatment,33 and
could play a role in carcinogenesis. By analogy, chronic inflammation associated
with persistent infection by Helicobacter pylori
strain CagA+ is a known risk factor for the development of gastric carcinoma
Chlamydia trachomatis antibodies and multiple
serotypes of C trachomatis have been detected in
women with several sex partners and in women with upper genital tract infections.20,26,36,37 The
presence of mixed infections implies that infection with one serotype does
not induce protective immunity against subsequent infection caused by another
serotype.36 Multiple exposures might increase
the risk of ultimately acquiring infections caused by the cancer-associated
serotypes, serotype G in particular (in this study, all but 1 case with antibodies
to ≥4 serotypes had antibodies to serotype G). On the other hand, mixed
serotypes are not common in patients with recurrent C trachomatis infections. Broadly reactive antigens of C trachomatis may simply result in humoral immune response against conserved cross-reactive
epitopes.31 Therefore, antibodies to multiple
serotypes discovered in patients with cervical SCC may also suggest chronic
infection by a single serotype. On the basis of this study we cannot distinguish
between these 2 possibilities.
Schlott et al38 detected C trachomatis DNA by in situ PCR in 40% of cervical carcinoma tissue
samples, but not in the carcinomatous cells.38
We found only 5% of the biopsy specimens positive for C
trachomatis DNA. The reason for this discrepancy is not known, but
assay-specific methodological differences are a plausible explanation. Futhermore,
the plasmid primers used for our PCR may not be optimal for the detection
of chronic C trachomatis infection. The target cells
for C trachomatis are endocervical glandular cells,39which might not be present in the diagnostic tumor
biopsy specimen. Finally, our previous seroepidemiologic6,8
and longitudinal DNA studies (K.-L. Wallin, PhD, F. Wiklund, MSc, T. Luostarinen,
MSc, et al, unpublished data, 2000) have indicated that exposure to C trachomatis takes place several years or even decades
before the diagnosis of cervical SCC. It is possible that the few tissue specimens
positive for C trachomatis DNA at the time of the
cancer diagnosis only represent the minimum number of C
trachomatis infections associated with cervical carcinoma.
Evidence based on longitudinal studies, such as the present study, is
always stronger than that based on cross-sectional surveys in which selection
bias is difficult to rule out. Our study was nested in a population-based,
well-defined cohort, suggesting that bias due to differential misclassification
was also highly unlikely. Use of serum banks and cancer registries with almost
100% reporting coverage provides an ideal setting for etiologic studies with
true cancer as the end point.
The development of SCC takes several years, probably decades. The link
between bacterial infections and carcinogenesis is not clear, but genetic
damage and neoplastic changes can be induced in vitro by coculturing cells
with activated inflammatory cells.40 Release
of nitric oxide occurs in C trachomatis infections.41 Recent studies have also shown that C trachomatis inhibits host cell apoptosis by specific mechanisms.42 In chronic chlamydial infections, these mechanisms
could initiate or promote cervical carcinogenesis. The serotype-specific differences,
and the fact that the risk was higher in women exposed to more than 1 serotype,
strengthen the evidence for the role of C trachomatis
in cervical carcinogenesis. It is tempting to speculate on the potential molecular
mechanisms explaining this association. Future studies should address the
question of whether there are any specific determinants related to serotype
G that may be directly or indirectly carcinogenic.