Article Text
Abstract
Background Increased duration of hormonal contraceptive (HC) use may be positively associated with the risk of invasive cervical cancer.
Methods This is a secondary analysis from the HPV Sentinel Surveillance Study. The authors examined the association between type-specific human papillomavirus (HPV) detection and current HC use among 7718 women attending 26 sexually transmitted disease, family planning and primary care clinics in the USA.
Results There was an association between HC use and HPV-16 detection (adjusted prevalence rate ratio 1.34 (95% CI 1.05 to 1.71) for oral contraceptive users and 1.41 (1.01 to 2.04) for depot-medroxyprogesterone acetate users); there was no association between HC use and detection of other HPV types or any HPV overall.
Conclusions Longitudinal studies are needed to better define this type-specific association and its clinical significance.
- Hormonal contraception
- HPV
- human papillomavirus
- Papanicolaou smear
- infectious disease
- treponemal infection
- Treponema pallidum
- syphilis serology
- law ethics
- notification
- chlamydia
- epidemiology
- cervical cancer
- self-sampling
- cervix
- epidemiology
- contraception
Statistics from Altmetric.com
- Hormonal contraception
- HPV
- human papillomavirus
- Papanicolaou smear
- infectious disease
- treponemal infection
- Treponema pallidum
- syphilis serology
- law ethics
- notification
- chlamydia
- epidemiology
- cervical cancer
- self-sampling
- cervix
- epidemiology
- contraception
Introduction
Increased duration of hormonal contraceptive (HC) use may be positively associated with the risk of invasive cervical cancer.1 2 However, reports of an association between HC use and detection of human papillomavirus (HPV) DNA have been inconsistent.3–7 Some of these inconsistencies may be explained by differences in study design (eg, detection of incident vs. prevalent infection). Most studies, however, did not report any HPV type-specific associations with HC use. Our goal was to examine the association between type-specific HPV detection and HC use.
Materials and methods
This is a secondary analysis from the cross-sectional HPV Sentinel Surveillance (HSS) Study whose details have been described elsewhere.8 Briefly, women from 26 clinics in six cities were enrolled in this cross-sectional study between January 2003 and December 2005. The primary goal of the study was to measure the burden of HPV infection and abnormal Papanicolaou (Pap) test results in the US cervical screening population by using broad sampling and a standardised protocol. Women aged 14–65 years were invited to participate if they were eligible for routine Pap screening. Exclusion criteria included a previous hysterectomy, a Pap test in the past 12 months, treatment of the cervix in the past 12 months, active menstruation and pregnancy. An administered questionnaire was used to collect data on demographics, sexual behaviours and other behaviours. Data on medical diagnoses (eg, other sexually transmitted infections) and Pap test results were abstracted from medical records.
Clinicians collected cervical specimens (either conventional or liquid based) for Pap testing. Cytopathologists at each of the six cities interpreted all cytology specimens by using the 2001 Bethesda Guidelines. An additional sample using the Digene Cervical Sampler (Digene, Gaithersburg, Maryland, USA) was collected and tested for HPV DNA using the prototype Roche Line probe assay (reagents provided by Roche Molecular Diagnostics, Alameda, California, USA) and sequencing as previously described. The line probe detected high-risk (HR) types (16, 18, 26, 31, 33, 35, 39, 45, 51, 52, 53, 56, 58, 59, 66, 67, 68, 69, 70, 73, 82, IS39) and low-risk (LR) types (6, 11, 40, 42, 54, 55, 61, 62, 64, 71, 72, 81, 83, 84, 89).9 We assessed the association of HC use with any HPV detection overall, with detection of only HR or LR types, with detection of two HPV clades, which contain common HR HPV types (clade A7, which includes HPV types 18, 39, 45, 59 and 68, and clade A9, which includes HPV types 16, 31, 33, 35, 52 and 58)10 and with four of the more common types associated with cervical cancer: HPV 16, 18, 31 and 45.11 The study was approved by human subjects committees at each of the participating sites and the Centers for Disease Control and Prevention (CDC). Funding for the study was provided by the CDC, which was involved in the conduct, design and reporting of the study. For this analysis, we included only those women who had documented information as reported on an interviewer-administered questionnaire about the current use of HC, either oral contraceptives (OCS) or depot-medroxyprogesterone acetate (DMPA), and those with valid HPV DNA test results.
The sample size calculation was based on the primary objectives of the parent study.8 Ninety-five per cent CIs are provided as a measure of the uncertainty surrounding each point estimate. The χ2 test was used to compare independent proportions. Because the outcomes of interest occurred frequently, we calculated the prevalence rate ratios (PRRs) instead of the ORs. Associations were assessed using log binomial regression models to generate PRR. Models were adjusted for the following variables: age, race, clinic type, age at first intercourse, total number of children, current pregnancy, lifetime number of sex partners, new partners in past 6 months, current tobacco use, past tobacco use, HIV status, condom use and reproductive tract infection (only for the Pap smear analysis). We assessed the statistical interaction between age and HC use as a proxy for the duration of HC use.
Results
Of the 9657 women included in the original study, 7718 (80%) had HC data available: 1490 (20.1%) were current OCS users and 478 (6.4%) currently used DMPA. No single variable had >5% missing data except for the number of children (table 1). The majority of women on HC were recruited from family planning clinics. Women taking HC were, in general, younger and more likely to be white; they reported less tobacco use, fewer sexual partners, older age at sexual debut, less condom use and fewer children. The demographic and clinical characteristics, stratified by HC use, are summarised in table 1.
The overall prevalence of any HPV type in this subset of women was 32.5%. There was no association between HC use (either OCS or DMPA) and the detection of any HPV DNA, any HR HPV DNA, any LR HPV DNA and any clade A7 or A9 HPV DNA (table 2). There was a significant association between HPV-16 detection and OCS (aPRR 1.34; 95% CI 1.05 to 1.71) and DMPA (1.41; 1.01 to 2.04) use after adjusting for several known confounders (table 2). Among women who had normal findings on their Pap smears, the associations between HPV-16 detection and combined oral contraceptive use remained significant (aPRR 1.41; 95% CI 1.05 to 1.89), but the association between HPV-16 and DMPA lost statistical significance (aPRR 1.44; 95% CI 0.78 to 1.96). There was no significant interaction between age and HC use (a proxy measure for the duration of HC use). There was no association between current HC use and dysplastic changes on Pap smear (table 2), even among the subset of women infected with HPV-16 (for OCS users, the PRR of dysplastic changes among HPV-16-infected persons was 1.13 (95% CI 0.90 to 1.40); for DMPA users, PRR 0.68 (95% CI 0.44 to 1.05)). There was an association between OCS use and the detection of atypical squamous cells of unclear significance on Pap smear. This association, however, was no longer statistically significant when excluding women with detectable HPV-16 DNA.
Discussion
The lack of an overall association between HC use and HPV detection is in agreement with several other large published studies.4 7 In a systematic review that included 19 epidemiological studies, there was no association found between HC use and the detection of any HPV types, including HR or LR HPV types overall.4 In addition, duration of HC use did not impact the probability of HPV detection. Similarly, in a pooled analysis of HPV surveys among 15 145 women conducted by the International Agency for Research on Cancer, HPV detection was similar in long-term OCS users and non-users.7 Neither of these studies evaluated specific HPV types. A longitudinal study of incident HPV infections among 1997 women reported an association between current OCS use and the diagnosis of any incident HPV infection. Although no type-specific association with HC use was reported, HPV-16 was the most frequently detected type accounting for 10.3% of all incident infections.5
In our study, although there was no overall association between HR or LR HPV types, we found an intriguing type-specific association between HPV-16 and HC use. HPV-16 is the most common HPV type that causes cervical cancer in women.11 There are several biological mechanisms that could explain an association between HPV-16 detection and HC use. Glucocorticoid-responsive elements (GREs) that regulate HPV viral transcription respond to progesterones by increasing the transcription of E6 and E7 genes of certain HPV types—including HPV-16 (reviewed by de Villiers12). However, not all HR HPV GREs respond to the same degree. These data suggest type specificity and provide a biological mechanism to account for the increased detection of HPV-16 and its greater oncogenic potential; they cannot, however, completely explain our findings as GREs from the HR-type HPV-18 and LR-type HPV-11 are also thought to respond to progesterone. In addition to increased viral replication, recent data also suggest that other biological mechanisms may be important. Sex hormones significantly alter the immune responses to HPV-16 virus-like particles.13 This raises the possibility that HC may also affect the host's susceptibility to HPV-16 infection. Other data suggest the possibility of progesterone-mediated persistence of HPV-16 infection thereby increasing the likelihood of its detection.14 Of note, there was no association between clade A9 (which includes HPV type 16) and HC use, suggesting that the association between HPV-16 and HC use is type specific and not clade specific. Clade specificities in GRE hormone responsiveness, immune responses and persistence, to our knowledge, have not been studied.
In our data, there was no short-term impact of HC use on cervical dysplasia as measured by Pap smear, even in the subset of women who had detectable HPV-16 DNA. This finding mirrors the results of two case control studies that assessed the impact of HC use on cervical intraepithelial neoplasia.15 Dysplastic changes, however, develop over time, so a cross-sectional design is not ideal to study this association. Several other studies, however, including a systematic review2 and a multicentre case control study1 suggest an association between HC use and the development of cervical cancer. Whether the increased detection of cervical cancer among HC users may be mediated, in part, by a type-specific interaction between HPV-16 (and possibly other HR HPV types) and HC is not known.
This study has several limitations. The cross-sectional design introduces the potential for selection and information biases. The large number of women recruited and the variety of settings from which they were recruited may help limit some of these. We had limited data on the type, timing and duration of HC use, so no causal link between HC use and type-specific HPV detection can be inferred. Despite our large sample size, the low prevalence of some HPV types may have limited the power of this study to detect other type-specific associations with HC use. However, the lack of an overall association between HC use and HR HPV overall, LR HPV overall and HPV clades (all of which had a higher prevalence than HPV-16) suggests that the type-specific association described with HPV-16 is not due to its higher prevalence. Finally, we cannot completely discount the possibility that our finding is the result of behavioural changes or chance alone.
Over 300 million women worldwide use HC. Defining the impact of HC use on HPV infection is critically important. This study suggests a type-specific association between HPV-16 detection and HC use. Longitudinal studies are needed to better define this association and determine its clinical significance.
Key messages
We detected a cross-sectional type-specific association between HPV-16 and hormonal contraception use.
There was no cross-sectional association between hormonal contraception use and Papanicolaou smear abnormalities.
Longitudinal studies are needed to better define these associations and determine their clinical impact.
References
Footnotes
Funding All funding for the HPV Sentinel Surveillance Study was provided by the Centers for Disease Control and Prevention. Additional funding from the National Institutes of Health (K23HD047395) was provided to KGG.
Competing interests Honoraria: KGG (Merck). Portions of the data were presented (abstract #P2.54) at the 18th International Society for Sexually Transmitted Diseases Research conference in London, 28 June–1 July 2009.
Ethics approval IRBs of all participating institutions.
Provenance and peer review Not commissioned; externally peer reviewed