Frequency and risk factors for prevalent, incident, and persistent genital carcinogenic human papillomavirus infection in sexually active women: community based cohort studyBMJ 2012; 344 doi: https://doi.org/10.1136/bmj.e4168 (Published 22 June 2012) Cite this as: BMJ 2012;344:e4168
- Pippa Oakeshott, reader in general practice1,
- Adamma Aghaizu, scientist epidemiology1,
- Fiona Reid, senior lecturer in medical statistics1,
- Rebecca Howell-Jones, scientist epidemiology2,
- Phillip E Hay, reader in genitourinary medicine3,
- S Tariq Sadiq, reader in sexual health/HIV medicine3,
- Charles J Lacey, professor of genitourinary medicine4,
- Simon Beddows, microbiologist5,
- Kate Soldan, epidemiologist2
- 1Division of Population Health Sciences, St George’s, University of London SW17 0RE, UK
- 2HIV and STI Department, Health Protection Agency, London NW9 5EQ
- 3Department of Genitourinary Medicine, St George’s, London
- 4Hull York Medical School, University of York, York YO10 5DD
- 5Virus Reference Department, Health Protection Agency, London
- Correspondence to: P Oakeshott
- Accepted 8 May 2012
Objective To investigate frequency and risk factors for prevalent, incident, and persistent carcinogenic human papillomavirus (HPV) in young women before the introduction of immunisation against HPV types 16 and 18 for schoolgirls.
Design Cohort study
Setting 20 London universities and further education colleges.
Participants 2185 sexually active female students, mean age 21 years (range 16-27), 38% from ethnic minorities, who took part in the POPI (prevention of pelvic infection) chlamydia screening trial in 2004-08 and who provided duplicate, self taken vaginal swabs and completed questionnaires at baseline. At follow-up, a median of 16 months later, 821 women (38%) returned repeat vaginal swabs by post. In 2009-10, stored samples were tested for HPV.
Results Samples from 404/2185 (18.5% (95% CI 16.9% to 20.2%)) of the cohort were positive for carcinogenic HPV at baseline, including 15.0% (327) positive for non-vaccine carcinogenic genotypes. Reporting two or more sexual partners in the previous year and concurrent Chlamydia trachomatis or bacterial vaginosis were independent risk factors for prevalent vaginal HPV infection. Infection with one or more new HPV types was found in 17.7% (145/821) of follow-up samples, giving an estimated annual incidence of carcinogenic HPV infection of 12.9% (95% CI 11.0% to 15.0%). Incident infection was more common in women reporting two or more partners in the previous year, aged<20, of black ethnicity, or with C trachomatis vaginosis at baseline. Multiple partners was the only independent risk factor for incident infection (adjusted relative risk 1.99 (95% CI 1.46 to 2.72)). Of 143 women with baseline carcinogenic HPV infection, 20 (14% (8.3% to 19.7%) had infection with the same carcinogenic HPV type(s) detected after 12-28 months. Of these women, 13 (65%) had redetected infection with HPV 16 or 18, and nine (45%) with non-vaccine carcinogenic HPV genotypes.
Conclusion In the first UK cohort study of carcinogenic HPV in young women in the community, multiple sexual partners was an independent predictor of both prevalent and incident infection. Infection with non-vaccine carcinogenic genotypes was common. Although current HPV vaccines offer partial cross protection against some non-vaccine carcinogenic HPV types, immunised women will still need cervical screening.
Each year around 500 000 women are diagnosed with invasive cervical cancer worldwide and 250 000 women die from it, mainly in developing countries.1 Persistent infection with a carcinogenic human papillomavirus (HPV) is a prerequisite for cervical cancer,2 and around 20% of women with one year persistence of HPV infection in the cervix will develop cervical intraepithelial neoplasia or cervical cancer in the next five years.3 HPV genotypes 16 and 18 are responsible for around 70% of cervical cancers, with the remaining 30% caused by other carcinogenic HPV types. HPV16 is the most carcinogenic type in terms of numbers of cases of cervical intraepithelial neoplasia or cervical cancer.4 5 Low risk genotypes 6 and 11 cause most cases of genital warts. Although HPV infections are the most common viral sexually transmitted infection, most infections are transient and asymptomatic, with about 70% of new infections resolving within a year.2 6 7
Immunisation against HPV types 16 and 18 protects against cervical intraepithelial neoplasia.8 9 Since 2008, the UK-wide HPV immunisation programme has been vaccinating adolescent schoolgirls against carcinogenic HPV16 and HPV18, but there is a dearth of UK baseline data on HPV infection in women from non-healthcare settings, especially those in some ethnic minorities who may have a higher risk of cervical cancer but lower uptake of immunisation and cervical screening.10 11 These data are essential to monitor the impact of the vaccination programme and to guide the ongoing introduction of HPV testing.
We used stored vaginal samples from women who took part in the POPI (prevention of pelvic infection) chlamydia screening trial,12 which was before the introduction of the immunisation programme, to investigate the frequency and risk factors for prevalent, incident, and persistent carcinogenic HPV infection in young women in the community.
The design of the POPI trial has been described elsewhere.12 Briefly, during 2004-06, 2529 female students were recruited from 20 London universities and further education colleges. Students were eligible to take part if they were aged ≤27 years, female, sexually experienced, not pregnant, and had not been tested for Chlamydia trachomatis infection in the previous three months. They were asked to complete a questionnaire and to provide two self-taken vaginal swabs.13 One swab was used for the chlamydia screening trial. The other was rolled over a glass slide for bacterial vaginosis analysis, placed in Aptima transport medium (Gen Probe) and stored at –80°C. Twelve months after recruitment, 94% of participants were followed up by questionnaire or medical record search.12 Those who agreed to repeat vaginal swabs were sent two swabs for return by post. One swab was tested for C trachomatis immediately and the result fed back to the participant, the other was stored. For the current study we analysed stored baseline samples (including a subset previously described14) and the follow-up samples, enabling repeat testing for HPV.
Laboratory testing for HPV
In 2009-10 the stored specimens from the baseline and follow-up vaginal swabs were tested for HPV by the Health Protection Agency. HPV testing was conducted anonymously; results were not fed back to participants. Samples were initially screened for HPV infection with the Digene Hybrid Capture 2 assay, which includes probes for high risk and low carcinogenic risk. The specimens that tested positive were subsequently genotyped using the Roche Linear Array HPV Genotyping assay.14 15 This test can detect 37 HPV types (6, 11, 16, 18, 26, 31, 33, 35, 39, 40, 42, 45, 51, 52, 53, 54, 55, 56, 58, 59, 61, 62, 64, 66, 67, 68, 69, 70, 71, 72, 73 (MM9), 81, 82 (MM4), 83 (MM7), 84 (MM8), IS39, and CP6108) and includes a β-globin probe to check for sample integrity to reduce the risk of false negatives. Of 323 samples that tested positive with the Hybrid Capture 2 assay but negative with the Linear Array assay, β-globin was not detected in two baseline samples and these were excluded. In line with the 2009 classification by the International Agency Research on Cancer,16 we defined 13 “probably carcinogenic” genotypes as high risk carcinogenic HPV types (16, 18, 31, 33, 35, 39, 45, 51, 52, 56, 58, 59, and 68).
We defined incident HPV infection as a positive swab at follow-up in women who were negative for the same carcinogenic HPV genotype at baseline. Infection with the same carcinogenic HPV genotype in both baseline and follow-up samples was classified as persistence or redetection of HPV infection, recognising that some of these cases could have been due to clearance and subsequent re-infection with the same HPV genotype.8
The sample size for the study was constrained by the 2529 women recruited to the POPI trial.12 We assumed the prevalence of HPV16 was 5%.4 Given that 28% of the cohort was of black ethnicity,12 this sample size would allow detection of a difference in the prevalence of HPV16 of 7% in black women compared with 4% in the remainder with 80% power and 5% significance.
We investigated demographic and behavioural risk factors often associated with sexually transmitted infections: young age, black ethnicity, smoking, multiple sexual partners, sexual debut at age <16 years, and use of oral contraception or condoms.6 13 17 18 As sexually transmitted infections often coexist, we also looked at concurrent Chlamydia trachomatis, Mycoplasma genitalium, Neisseria gonorrhoeae, or bacterial vaginosis.
We used data on the baseline samples for analysis of predictors of prevalent carcinogenic HPV infection and used data for women who returned follow-up samples to analyse predictors of incident infection and persistent or redetected infection. We examined the relation between possible risk factors and prevalent carcinogenic HPV infection at baseline, and incident or persistent or redetected HPV at follow-up, using binomial regression with a log link in Stata, version 10. Results are presented as relative risks. For prevalent and incident infections, multivariate analyses were conducted which included all variables found to be statistically significant in the univariate analyses, plus variables that were non-significant but have previously been shown to be related to HPV infection (age <20, age <16 at sexual debut, and smoking6 17 18). As the number of persistent or redetected infections was small, we performed only univariate analyses for the eight risk factors found to be associated with prevalent or incident infection.
We estimated the annual incidence of HPV infection by dividing the number of incident cases by the total person years of follow-up from date of recruitment to date of providing the repeat vaginal sample (range 11-32 months). We then expressed the rate per 100 person years as the percentage annual incidence. Calculation of the true incidence for new infections would have required information on the date of infection, which was not available. We calculated rates of persistence or redetection for each carcinogenic genotype for HPV positive women at baseline who returned repeat samples by dividing the number of women with the same HPV genotype at both baseline and follow-up by the number with that genotype at baseline.
Of the 2529 women recruited to the trial, 2185 (86%) provided duplicate vaginal samples at baseline that were suitable for HPV analysis (see figure⇓). Of these women, 821 (38%) returned repeat postal samples after 11-32 months (median 16 months). Compared with non-responders, the women who returned repeat samples were slightly older and less likely to be of black ethnicity, or to smoke, use condoms, or have bacterial vaginosis at baseline (see appendix table 1 on bmj.com).
Prevalence of carcinogenic HPV infection at baseline
Of the 2185 women, 404 (18.5% (95% CI 16.9% to 20.2%)) were positive for carcinogenic HPV at baseline (table 1⇓). HPV16 was the most common carcinogenic HPV infection (prevalence 5.4% (4.5% to 6.4%)). However, 327 women (15%) had infection with at least one non-vaccine carcinogenic HPV type and 144 (7%) had multiple carcinogenic HPV infections. Of the 51 women infected with HPV6 and nine women infected with HPV11, only two (3.3%) gave a history of genital warts.
Predictors of prevalent carcinogenic HPV infection
Carcinogenic HPV infection was more common among women of black ethnicity, those who smoked, those reporting two or more sexual partners in the previous 12 months, those who reported using condoms (no data available on frequency of use), and those with concurrent C trachomatis, M genitalium, N gonorrhoeae, or bacterial vaginosis (table 2⇓). Multivariate analysis showed that having multiple partners in the previous 12 months and co-infection with C trachomatis or concurrent bacterial vaginosis were independent predictors of carcinogenic HPV infection.
Incidence of carcinogenic HPV infection
Of the 821 women who provided follow-up samples, 145 (17.7% (15.1% to 20.4%)) were infected with one or more new HPV genotypes (table 1⇑ and figure⇑). Taking into account the timing of the follow-up sample (median 16 months after recruitment), the estimated annual incidence was 12.9% (11.0% to 15.0%) for any carcinogenic HPV infection and 9.8% (8.1% to 11.7%) for non-vaccine genotypes. Predictors of incident infection were age <20 years, black ethnicity, reporting two or more sexual partners in the previous 12 months, reported condom use, and C trachomatis infection at baseline (table 3⇓). Having two or more partners was the only independent predictor of incident carcinogenic HPV infection.
Persistence or redetection of carcinogenic HPV infection of specific type
Of the 821 women returning follow-up samples, 143 (17.4%) had one or more carcinogenic HPV infections at baseline, and 20/821 (2.4% (1.4% to 3.5%)) had infection with at least one of the same genotypes redetected after 12-28 months. Thus one in seven women (20/143, 14.0% (8.3% to 19.7%)) with baseline infection had either not cleared their infection or been reinfected with the same genotype (table 1⇑). Of these 20 women, 13 (65%) had persistent or redetected HPV16 or HPV18 infection and nine (45%) had a non-vaccine HPV type. Persistence or redetection of specific carcinogenic HPV types was not significantly associated with sexual behaviour, demography, smoking, or co-infection but the numbers were small (appendix table 2 on bmj.com).
Vaccine HPV types (HPV16 and HPV18)
Rates of prevalent, incident, and persistent or redetected infection with HPV16 or HPV18, or both, are shown in table 4⇓. Around one in five women with HPV16 or HPV18 at baseline had persistent or redetected infection after a median of 16 months follow-up. Infections with vaccine HPV types were more likely to persist or be redetected than infections with non-vaccine HPV types: 21% (13/63) versus 9% (9/104).
In this community based cohort of multiethnic female students, the prevalence and annual incidence of carcinogenic HPV were 18% and 13%, and reporting multiple sexual partners in the previous 12 months was an independent predictor of both prevalent and incident infection. Among women who returned follow-up samples after a median of 16 months, 14% of those with baseline infection (that is, 2% overall) had persistent or redetected HPV infection with the same carcinogenic genotype. Nearly half of these women had persistent or redetected infection with non-vaccine HPV genotypes.
Strengths and weaknesses
This is the first UK cohort study of carcinogenic HPV infection in women recruited at educational institutions rather than from healthcare facilities. It provides useful baseline data on HPV infection in England before the introduction of HPV immunisation. It enabled some investigation of the association of carcinogenic HPV infection with concurrent C trachomatis, M genitalium, N gonorrhoeae, or bacterial vaginosis. More than a third of the women returning follow-up samples were teenagers, many from ethnic minorities. Although black ethnicity was associated with prevalent and incident carcinogenic HPV infection in univariate analysis, it was not a significant independent risk factor for infection. Finally the study provides new information on demographics, smoking, and sexual behaviour related to carcinogenic HPV infection in England.
The main weakness is the study was not population based. This limits generalisability of our findings. However, these data may be the best currently available from a community based cohort of young women in the UK. A recent Scottish population based survey requesting 5500 unscreened young women aged over 21 to take vaginal swabs and return them by post for HPV testing had a 13% response rate.19
We may have underestimated the annual incidence of carcinogenic HPV infection for three reasons. Only 38% of the cohort returned follow-up postal samples so data on incidence (and on redetection or persistence) are based on a self selected, probably lower risk group as they were older and less likely to smoke or to be from an ethnic minority. Secondly, we used length of follow-up as a proxy for time to infection, which will have overestimated the time at risk and underestimated the incidence. Thirdly, as median duration of HPV infection is around eight months,6 we probably missed some women who acquired and cleared a new HPV infection in the median 16 month interval between testing at baseline and at follow-up swab. Also we cannot be sure that some cases of apparent HPV persistence were not clearance and reinfection with the same genotype.
The sample size was restricted to 2185 women who took part in the POPI trial and 821 who participated in follow-up, limiting the power to identify and adjust for risk factors. Small numbers of women with persistent or redetected infection meant that risk factors for this could not be robustly evaluated. However, other cohort studies were of similar size.4 6 17 Although self collected vaginal swabs are reliable for detection of HPV,20 HPV types in the vagina may differ from those in the cervix. In the samples that screened positive for HPV infection but were negative in the HPV genotyping assay there were some indications of reduced sensitivity (notwithstanding the high rate of β-globin positivity indicating sample integrity), probably due to the long storage time.14 Finally, findings may not be applicable to different populations such as those in developing countries or women attending genitourinary, family planning, or hospital clinics.
Comparison with other studies
The most common carcinogenic HPV genotypes we found were 16, 18, 51, 52, and 59, which is similar to findings from other UK studies4 14 15 21 and multinational incidence studies in men.22 As in reports from clinic based studies, risk factors for prevalent and incident carcinogenic HPV infection were similar to those for other sexually transmitted infections: multiple sexual partners,6 17 18 young age,6 ethnic minority background,6 and co-infection.17 The unexpected association of prevalent HPV with reported condom use may be because condoms are more likely to be used, but not consistently, by those at higher risk. Others have also found reported condom use was not protective against HPV infection,18 although consistent condom use may be.23
However it is persistent carcinogenic HPV infection which explains virtually all cases of cervical cancer.1 Ho and colleagues’ finding that 28% (7/25) of women with HPV16 at baseline had persistent HPV16 infection for 24 months6 is similar to our redetection rate of HPV16 of 24% (11/46); and, as in our study, persistent infection for six months was not associated with smoking, but numbers were small.6 Another report from a selected population of women with equivocal or mildly abnormal cytology also suggested that six month persistence of HPV infection was not associated with smoking except possibly in women smoking more than 20 cigarettes daily.24
Implications of study
Many women diagnosed with carcinogenic HPV infection of the cervix are worried about the risk of developing cervical cancer. Although they can be reassured that most new infections clear spontaneously, we found that one in seven women with carcinogenic HPV infection at baseline had persistent or redetected infection for up to 28 months, nearly half of them with carcinogenic genotypes not covered by current HPV vaccines. In September 2012 the UK-wide HPV vaccination programme will be switching from the bivalent vaccine targeting HPV types 16 and 18 to the quadrivalent vaccine, which also targets HPV types 6 and 11.25 Both vaccines provide some cross protection against cervical intraepithelial neoplasia associated with carcinogenic HPV types 31, 33, 45, 52, and 58,9 26 27 though cross protection seems stronger with the bivalent vaccine.9 However, since around 20-30% of cervical cancers in the UK and many persistent infections are due to non-vaccine genotypes, choice of vaccine may need to be re-evaluated when a second generation broad spectrum HPV vaccine including these additional five carcinogenic genotypes becomes available. Meanwhile, both vaccinated and unvaccinated women should be informed of the importance of continued cervical cytology and/or HPV testing.1 26
New English guidelines recommend that cervical samples from women with borderline or mild dyskaryosis are analysed for carcinogenic HPV infection. Women who test positive will be referred for colposcopy. Those who are negative will be returned to routine recall. But policy makers need more evidence on which women are most at risk of carcinogenic HPV, which HPV genotypes should be included in assays used for cervical screening, and appropriate intervals between tests. Our findings may be used to inform modelling of the possible impact of HPV vaccination in the UK, including changes in the epidemiology of non-vaccine carcinogenic HPV types, and the development of cervical screening strategies.28 5
What is already known on this topic
Among women not immunised against HPV, around 20% of those with persistent carcinogenic HPV infection of the cervix for a year develop cervical intraepithelial neoplasia or cancer over the next five years
Immunisation against HPV types 16 and 18 can prevent cervical intraepithelial neoplasia due to these genotypes and gives partial protection against some other carcinogenic HPV genotypes
There are no UK data on risk factors for incident or persistent carcinogenic HPV in young women in the community
What this study adds
Women reporting multiple sexual partners in the previous year were at highest risk of incident carcinogenic HPV infection
Fourteen per cent of women with a baseline carcinogenic HPV infection had genital infection with the same carcinogenic HPV genotype redetected after 12-28 months, nearly half of them with genotypes not targeted by current HPV vaccines
Findings highlight the importance of continued cervical cytology and/or HPV testing for both vaccinated and unvaccinated women
Cite this as: BMJ 2012;344:e4168
We thank Natasha de Silva, Jeremy Anton, and Nitisha Pyndiah for organising and conducting the HPV testing. We thank Dr Harry Mallinson and Liverpool Aintree Laboratory for doing the chlamydia tests, and Dr Jorgen Jensen and Statens Serum Institut for doing the M genitalium tests.
Contributors: PO, AA, PEH, and FR designed the study. SB was responsible for the HPV testing, AA led the analysis with input from FR, RHJ, KS, and PO. PO wrote the first draft of the manuscript, and all authors contributed to and approved the final version.
Funding: BUPA foundation grant TBF-08-025 and Medical Research Council grant 80280. Gen-Probe provided the Aptima test kits. The HPV testing was supported by a grant from GlaxoSmithKline (study No EPI-HPV-109903). RHJ was funded by the Policy Research Programme in the Department of Health, UK (grant reference No 039/030). PO, STS, and PEH are members of the eSTI2 consortium funded by the UK Clinical Research Collaboration. The views expressed in the publication are those of the authors and not necessarily those of the Department of Health or other funders.
Competing interests: All authors have completed the ICMJE uniform disclosure form at www.icmje.org/coi_disclosure.pdf (available on request from the corresponding author) and declare: no support from any organisation for the submitted work; no financial relationships with any organisations that might have an interest in the submitted work in the previous 3 years; no other relationships or activities that could appear to have influenced the submitted work.
Ethical review: Bromley research ethics committee reference 07/Q0705/16.
Data sharing: No additional data available.
This is an open-access article distributed under the terms of the Creative Commons Attribution Non-commercial License, which permits use, distribution, and reproduction in any medium, provided the original work is properly cited, the use is non commercial and is otherwise in compliance with the license. See: http://creativecommons.org/licenses/by-nc/2.0/ and http://creativecommons.org/licenses/by-nc/2.0/legalcode.