Fukuchi, Emily MPH*; Sawaya, George F. MD*; Chirenje, Michael MD†; Magure, Tsitsi MD†; Tuveson, Jennifer BA*; Ma, Yong MS‡; Shiboski, Stephen PHD‡; Da Costa, Maria MSc§; Palefsky, Joel MD§; Moscicki, Anna-Barbara MD∥; Makunike-Mutasa, Rudo MD¶; Chipato, Tsungai MBCHB†; Smith-McCune, Karen K. MD, PHD*
CERVICAL CANCER IS THE MOST common neoplasm among women in Zimbabwe, with age-adjusted incidence rates estimated at 52.1 per 100,000 women and age-adjusted mortality rates of 43.1%.1 Of the 40 known genital oncogenic human papillomavirus (HPV) types, 15 are associated with cervical cancer.2–5 To date, there has been sparse data describing the type-specific prevalence of HPV in women in sub-Saharan Africa. Longitudinal studies of young women in North America and Europe have reported cumulative HPV incidence rates of approximately 35% to 50% within 3 years, with HPV16 being the most common type in most studies.6–11 Although most HPV infections are transient and resolve spontaneously, oncogenic types are more likely to persist and play a key role in carcinogenesis.12,13 Prospective cohort studies that examine the dynamic nature of HPV infections over time will aid in identifying independent risk factors associated with viral persistence, thereby assisting in the identification of modifiable risk factors that may have a role in prevention of cervical cancer.
We performed serial HPV testing in a cohort of human immunodeficiency virus (HIV)-negative Zimbabwean women enrolled in a randomized trial to assess the efficacy of diaphragm plus lubricant gel provision on HIV14 and HPV incidence.15 Our study is a secondary analysis of this cohort to examine type-specific HPV prevalence, type-specific HPV incidence, and type-specific HPV persistence over an average of 21 months of follow-up. We also determined demographic, behavioral, and biologic/clinical features of participants at enrollment that predicted HPV incidence and persistence.
MATERIALS AND METHODS
Participants were 2040 HIV-negative Zimbabwean women enrolled in a multicentered (Zimbabwe and South Africa), open-label, randomized trial assessing the effect of providing a latex diaphragm with a lubricant gel along with male condoms compared with male condom provision alone on HIV seroincidence; details and results of this trial, the Methods for Improving Reproductive Health in Africa Study (ClinicalTrials.gov number NCT00211459), are presented elsewhere.14 Women who enrolled in Methods for Improving Reproductive Health in Africa Study at the Harare, Zimbabwe site after February 2004 were asked to participate in a substudy to determine the effect of diaphragm and gel provision on the incidence and clearance of HPV infection. Enrollment was ongoing from February 2004 to September 2005. Those who agreed underwent a separate consent process; 2089 women were offered participation and 2040 women (97.6%) consented. Results of the primary analysis demonstrated that diaphragm plus lubricant gel provision in addition to condoms had no overall effect on HPV incidence or clearance;15 therefore, data from the 2 study arms were pooled for the secondary analysis presented here. The study protocol was reviewed and approved by the University of California, San Francisco Committee on Human Research, and the Medical Research Council of Zimbabwe, and the Medicines Control Authority of Zimbabwe.
Participants were followed quarterly from enrollment to September 2006. At enrollment, a baseline questionnaire on demographics and sexual behavior was administered, and participants were tested for serologic evidence of HIV and herpes simplex virus (HSV)-2; a urine specimen was obtained for PCR testing for Neisseria gonorrhoeae, Chlamydia trachomatis, and Trichomonas vaginalis, and a blood sample for syphilis testing using methods described elsewhere.14 A pelvic exam was performed, cervical cytology was obtained, and participants were examined for genital warts. Cervical cytology was interpreted locally in Zimbabwe (R. M.) and reported using the Bethesda system.16 During the trial, 88 women (4.3%) seroconverted to HIV and their outcomes were included in all applicable analyses.
Methods for Sample Collection and Testing
After cytology samples were taken at the enrollment visit, the cervix and endocervix were swabbed for HPV DNA with a polyester swab under direct visualization and stored as previously described.15 At the 12-month follow-up visit and at study exit (range 12–24 months), clinician-collected swabs were collected under identical conditions. Self-collected vaginal swabs were also obtained at 3-month intervals but were not included in this analysis.
HPV testing was performed using MY09/MY11 consensus HPV L1 primers and primers for amplification of the human β-globin gene as previously described.15 Samples that were negative for β-globin were considered inadequate and were not included in analyses. Specimens that tested positive for the consensus HPV sequence were further studied to determine the specific HPV type. The aforementioned procedure was repeated using probes specific to 29 different HPV types (6, 11, 16, 18, 26, 31, 32, 33, 35, 39, 40, 45, 51, 52, 53, 54, 55, 56, 58, 59, 61, 66, 68, 69, 70, 73, 83, 84, 82v), and the following 10 HPV types together in a probe mixture defined as “mix”: HPV 2, 13, 34, 42, 57, 62, 64, 67, 72, and 82. Although HPV55 is now considered to be a subtype of HPV44,17 we will continue to reference HPV55 as an individual type. Specimens that tested positive for the HPV consensus sequence, but negative for the 39 specific types were described as having “untyped” HPV.
Type-specific HPV prevalence for 29 types with a 95% confidence interval (CI) was calculated among all women with evaluable samples at enrollment. Type-specific HPV incidence was defined by detection of any new HPV type at the 12-month follow-up visit that was not identified at the enrollment visit; we included data from clinician-collected swabs during a window period of 11 to 16 months after enrollment. Type-specific persistence was defined by 2 consecutive positive tests for the same HPV type(s). Type-specific infections were excluded from this analysis if 1 of 2 consecutive tests were β-globin negative. We defined 3 categories of persistence: (a) prevalent persistent infections, defined as 2 consecutive type-specific positive HPV tests at enrollment and at least the 12-month follow-up visit; (b) incident persistent infections, defined as 2 consecutive type-specific positive HPV tests at the 12-month visit and exit visit, after a negative test at enrollment; and, (c) overall persistence, calculated by combining the incident and prevalent persistent infections. We excluded untyped HPV and mix for all type-specific persistence analyses. In evaluating type-specific prevalence, incidence, and persistence, a woman could contribute to multiple analyses for individual types if she had multiple HPV types detected at 1 visit. For example, a woman HPV-negative at enrollment, who subsequently tested positive for HPV16 and 18 at the 12-month visit, would be classified as having an incident infection for HPV18 and an incident infection for HPV16.
Point prevalence of infections with multiple HPV types was calculated for women testing positive for more than 1 HPV type at enrollment and 12-month visits. Unconditional multiple logistic regression was used to estimate adjusted odd ratios (ORs), with 95% CIs for potential risk factors for prevalent, incident, and persistent HPV infections. All risk factors were those assessed at the enrollment visit only. Twenty-two variables were used in bivariate models to construct the multivariate models. Final models included variables that were statistically significant in bivariate models (P ≤0.10) and were adjusted for randomization arm (comparison group consisted of women randomized to the control arm). For the regression analysis we defined the following 3 outcomes: (a) prevalence, defined as the detection of any HPV at enrollment; (b) incidence at 12 months, defined as the detection of any new HPV type not present at enrollment; (c) persistence, defined as the detection of the same HPV type(s) on 1 or more consecutive visits. The comparison group for type-specific prevalence consisted of women HPV-negative at enrollment. The comparison group for incident infection at 12 month consisted of women who did not have an incident infection. We defined persistence as a dichotomous outcome for women who had at least 1 positive result. Therefore, the comparison group for overall type-specific persistence consisted of women who had only 1 type-specific positive test out of 3 possible tests during follow-up. The longitudinal nature of this cohort study allowed us to assess the distinction between predictors of incident infections and persistent infections.
Most participants (94%) completed the trial, and the median follow-up period was 21 months (range 12–24 months). Table 1presents the distribution of sociodemographic and HPV risk factor variables obtained from a questionnaire administered at enrollment. The mean age of study participants at enrollment was 27 years (range 18–49 years). The majority of women reported living with a regular partner. The mean number of lifetime sexual partners was low. Few participants were smokers. Although only 7% of women tested positive for a sexually transmitted infection (STI) (chlamydia, gonorrhea, T. vaginalis, or syphilis), approximately 50% of women had serologic evidence of prior exposure to HSV-2. Of note, approximately two-thirds of women reported having a high-risk partner.
Prevalent HPV Infections
HPV prevalence at enrollment was based on a total of 1987 women. We excluded 53 women (2.6%) from this analysis because of inadequate samples for HPV testing (specimen negative for β- globin). The overall prevalence of HPV infection at enrollment was 24.5% (Table 2); 16.1% of women had oncogenic HPV infections.
Among the prevalent infections, HPV58 was the most common oncogenic type with 5.0% of women testing positive at enrollment, followed by HPV16 and 18 with 4.7% and 2.3% of women testing positive, respectively. HPV70 was the most common nononcogenic type with 2.4% of women testing positive. HPV6 and 11 infections were rare in this cohort with 0.7% and 0.2% of women testing positive, respectively.
Incident HPV Infections
We excluded 506 women due to a β-globin negative test at enrollment (n = 49), a β-globin negative test at 12 months (n = 47), or a clinician-collected swab obtained before 11 months or after 16 months (n = 294), untyped HPV at enrollment and follow-up (n = 9), and no postenrollment test due to loss to follow-up and/or withdrawal (n = 107). Type-specific incidence of any new HPV infection 12 months after enrollment was 23.3% (Table 3); 11.4% of women had an incident oncogenic HPV infection; 14.6% had an incident nononcogenic HPV infection. Among the incident infections, HPV58 was the most common oncogenic type (incident risk 2.4%), followed by HPV16 (2.1%). HPV70 was the most common nononcogenic type (1.7%).
Infections With Multiple HPV Types
At enrollment and 12 months, 6.4% and 7.0% of women were infected with more than 1 HPV type, respectively; the majority of these women were infected with 2 HPV types (approximately 70%). Among 125 women who were infected with more than 1 HPV type at enrollment, 48.8% had 2 or more oncogenic types. The most common HPV types found in multiple-type infections were HPV Types 58, 16, 70, 18, 53, and 33.
Persistent HPV Infections
At enrollment, there were 601 evaluable type-specific infections. At the 12-month follow-up visit, 161 (26.8%) type-specific infections persisted and were considered prevalent persistent infections. There were 425 evaluable type-specific infections at the 12-month visit. At the exit visit (median 21 months, range 12–24 months), 141 (33.2%) type-specific infections persisted and were considered incident persistent infections. Overall, 269 women had a persistent infection and 29.8% of all infections persisted (Table 4). In all cases, persistence of oncogenic types (37.3%, 95% CI, 33.2–41.6) was significantly higher than that of nononcogenic types (21.9%, 95% CI, 18.4–25.9).
Baseline Risk Factors for Prevalent, Incident, and Persistent HPV Infections
In univariate analyses, enrollment characteristics that were independently associated with a prevalent HPV infection were abnormal enrollment cytology, greater than 3 live births, having sex greater than 3 times a week, reporting condom use always or sometimes (compared to never), greater than 2 cesarean-sections, having a high-risk sexual partner, reporting more than 1 lifetime sexual partner, and HSV-2 seropositivity. Older age was protective. In the multivariate analysis, enrollment characteristics that were independently associated with a baseline prevalent HPV infection were abnormal enrollment cytology (OR 2.87, 95% CI, 2.23–3.69), having a high-risk sexual partner (OR 1.67, 95% CI, 1.30–2.14), reporting more than 1 lifetime sexual partner (OR 1.61, 95% CI, 1.25–2.06), and HSV-2 seropositivity (OR 1.36, 95% CI, 1.09–1.71). Older age was protective (OR 0.76, 95% CI, 0.70–0.84) and showed a linear relationship between decreasing age and prevalent HPV infection (data not shown).
In the multivariate analysis, the enrollment characteristics that were independently associated with an incident HPV infection at 12 months of follow-up were having more than 1 lifetime sexual partner (OR 2.05, 95% CI, 1.55–2.70), testing positive for an STI (chlamydia, gonorrhea, T. vaginalis, or syphilis) at enrollment (OR 1.66, 95% CI, 1.07–2.59), HSV-2 seropositivity (OR 1.40, 95% CI, 1.09–1.81), reporting first sex at age greater than 16 years (OR 1.61, 95% CI, 1.00–2.61), being HPV positive at enrollment for a different HPV type (OR 1.37, 95% C,I 1.04–1.80), and reporting using condoms only sometimes within 3 months before enrollment (compared to using condoms never) (OR 1.35, 95% CI, 1.00–1.81) (Table 5). Older age was protective (OR 0.79, 95% CI, 0.72–0.87); and showed a linear relationship between decreasing age and incident HPV infection (data not shown).
In the multivariate analysis, the enrollment characteristics that were independently associated with an overall type-specific persistent HPV infection were abnormal cytology (OR 2.05, 95% CI, 1.53–2.75), more than 1 lifetime sexual partner (OR 1.35, 95% CI, 1.10–1.98), and having a high-risk partner (OR 1.35, 95% CI, 1.10–1.98) (Table 5). Older age was protective (OR 0.90, 95% CI, 0.82–1.00) and showed a linear relationship between decreasing age and persistent HPV infection (data not shown).
In this large community-based prospective cohort of HIV-negative Zimbabwean women, we found a relatively high frequency of HPV58 (5.03% compared with 0.6% in North America and 0.3% in Europe).1 In addition, HPV58 was the most common HPV type identified in this cohort. Historically, HPV16 and 52 have been reported as the most common types in the African continent, although data from Africa are limited by small numbers of samples analyzed.1 HPV58 is associated with 3.3% of cervical cancers globally and 1.5% of cancers in Africa.1 A high prevalence of HPV58 has also been reported in several Eastern Asian and South-Eastern Asian countries, and in Asia, HPV58 is associated with more cases of cervical cancers than any other HPV type except for HPV16 and 18.1 Moreover, HPV58 has been implicated as an important cause of cervical neoplasia in a cohort of Chinese women (OR 3.98, 95% CI, 1.22–14.35) and increasing HPV58 prevalence was associated with increasing cervical lesion severity.18 HPV58 has also been associated with an increased risk of preinvasive cervical lesions in the setting of infection with multiple HPV types.19,20 The unexpectedly high prevalence of HPV58 described here provides a unique opportunity to study the contribution of HPV58 to the burden of cervical cancer in Zimbabwe.
Cross-sectional studies of HPV prevalence are unable to distinguish newly incident from persistent HPV infections. Because of the prospective nature of our study, we were able to make this distinction and thereby identify unique risk factors for each of these states. For example, onset of sexual activity after the age of 16 was a predictor of incident but not persistent infection. Testing positive for HSV-2 by serology or having an STI at enrollment was associated with an increased risk of an incident, but not persistent, HPV infection. The association with HSV-2 has been described by others,10 and is especially significant given known correlations between HSV-2 and the development of cervical cancer.21,22 The association between HSV-2 seropositivity or STI-positivity at enrollment and incident HPV infection could simply be a marker for high-risk behavior or may reflect other confounding factors. Alternatively, local disruption to the cervical epithelium from HSV recurrence may facilitate exposure of the HPV virions to the basal cell layer.22 This finding raises the possibility that HSV-2 and STI control measures may have a beneficial effect in preventing HPV infection.
Women who reported having a high-risk partner at enrollment were significantly more likely to have a persistent type-specific HPV infection than those who reported not having a high-risk partner. This novel finding likely represents continued reinfection of the women by the male partner over time. Given that we were limited to the parameters of partner risk as defined in the main trial, we were unable to assess other characteristics of the male partners that might affect his predisposition to harbor or transmit HPV, such as HIV infection. Further investigation of the characteristics of a woman’s partner may identify additional risk factors for a woman’s overall risk of persistent HPV infection.
Women who reported infrequent condom use in the 3 months before study entry were at an increased risk of acquiring a new HPV infection compared with women who reported never using condoms. This seemingly paradoxical finding is consistent with a prior study reporting that irregular condom use was associated with increased risk of gonorrhea and chlamydia infection compared to no condom use (OR 1.44, 95% CI, 1.06–1.99).23 These findings suggest that women who report inconsistent condom use may perceive their partners to be high-risk compared with women who find no need to use condoms. Therefore, it is important for condom distribution programs to stress consistent condom use to prevent STIs.
Our study had several strengths, including the large sample size, high retention rate, prospective nature, and the comprehensive information available on risk factors. Our study had several limitations, many of which are inherent to observational studies. Because the first follow-up visit occurred 12 months after the enrollment visit, it is possible that infections of short duration could have been missed. We acknowledge that we cannot distinguish between a true incident infection and a reactivation of a latent infection using the methods employed in this study. Despite adjusting for confounders we could define and measure at enrollment, our results may be due to unmeasured and unknown confounders. Additionally, we adjusted only for potential confounders identified at enrollment; these may well have changed over time and affected our findings. Also, enrollment predictors may be more relevant to HPV infections acquired earlier in the study and causal associations may be limited. Although controlling for the study arm did not affect our results, the fact that our findings are derived from combining data from the 2 study arms may limit its interpretation as a natural history study. Because of the small sample sizes of some subgroup analyses, we may have had inadequate power to demonstrate important associations.
We have identified having a high-risk partner as a potentially modifiable predictor of persistent HPV infection and HSV-2 seropositivity as predictor of incident HPV infection. The finding that HPV58 is the most common HPV type in Zimbabwean women will likely have implications for vaccine development in programs targeting African women.
1. Castellssague X, de Sanjose S, Aguado T, et al. HPV and Cervical Cancer in the World 2007 Report. WHO/ICO Information Centre on HPV and Cervical Cancer (HPV Information Centre); 2007.
2. Walboomers JM, Jacobs MV, Manos MM, et al. Human papillomavirus is a necessary cause of invasive cervical cancer worldwide. J Pathol 1999; 189:12–19.
3. Schiffman MH, Bauer HM, Hoover RN, et al. Epidemiologic evidence showing that human papillomavirus infection causes most cervical intraepithelial neoplasia. J Natl Cancer Inst 1993; 85:958–964.
4. Bosch FX, de Sanjose S. Chapter 1: Human papillomavirus and cervical cancer-burden and assessment of causality. J Natl Cancer Inst Monogr 2003:3–13.
5. Munoz N, Bosch FX, de Sanjose S, et al. Epidemiologic classification of human papillomavirus types associated with cervical cancer. N Engl J Med 2003; 348:518–527.
6. Woodman CB, Collins S, Winter H, et al. Natural history of cervical human papillomavirus infection in young women: A longitudinal cohort study. Lancet 2001; 357:1831–1836.
7. Winer RL, Feng Q, Hughes JP, et al. Risk of female human papillmavirus acquisition associated with first male sex partner. J Infect Dis 2008; 197:279–282.
8. Richardson H, Kelsall G, Tellier P, et al. The natural history of type-specific human papillomavirus infections in female university students. Cancer Epidemiol Biomarkers Prev 2003; 12:485–490.
9. Ho GY, Bierman R, Beardsley L, et al. RD. Natural history of cervicovaginal papillomavirus infection in young women. N Engl J Med 1998; 338:423–428.
10. Moscicki AB, Hills N, Shiboski S, et al. Risks for incident human papillomavirus infection and low-grade squamous intraepithelial lesion development in young females. JAMA 2001; 285:2995–3002.
11. Rodriguez AC, Burk R, Herrero R, et al. The natural history of human papillomavirus infection and cervical intraepithelial neoplasia among young women in the Guanacaste cohort shortly after initiation of sexual life. Sex Transm Dis 2007; 34:494–502.
12. Ylitalo N, Josefsson A, Melbye M, et al. A prospective study showing long-term infection with human papillomavirus 16 before the development of cervical carcinoma in situ. Cancer Res 2000; 60:6027–6032.
13. Kjaer SK, van den Brule AJ, Paull G, et al. Type specific persistence of high risk human papillomavirus (HPV) as indicator of high grade cervical squamous intraepithelial lesions in young women: Population based prospective follow-up study. BMJ 2002; 325:572.
14. Padian NS, van der Straten A, Ramjee G, et al. Diaphragm and lubricant gel for prevention of HIV acquisition in southern African women: A randomised controlled trial. Lancet 2007; 370:251–261.
15. Sawaya GF CM, Magure T, Tuveson J, et al. Effect of diaphragm and lubricant gel provision on human papillomavirus infection among women provided with condoms. Obstet Gynecol 2008; 112:990–997.
16. Solomon D, Davey D, Kurman R, et al. The 2001 Bethesda System: terminology for reporting results of cervical cytology. JAMA 2002; 287:2114–2119.
17. de Villiers EM, Fauquet C, Broker TR, et al. Classification of papillomaviruses. Virology 2004; 324:17–27.
18. Chan PK, Li WH, Chan MY, et al. High prevalence of human papillomavirus type 58 in Chinese women with cervical cancer and precancerous lesions. J Med Virol 1999; 59:232–238.
19. Trottier H, Mahmud S, Costa MC, et al. Human papillomavirus infections with multiple types and risk of cervical neoplasia. Cancer Epidemiol Biomarkers Prev 2006; 15:1274–1280.
20. Fife KH, Cramer HM, Schroeder JM, et al. Detection of multiple human papillomavirus types in the lower genital tract correlates with cervical dysplasia. J Med Virol 2001; 64: 550–559.
21. Daling JR, Madeleine MM, McKnight B, et al. The relationship of human papillomavirus-related cervical tumors to cigarette smoking, oral contraceptive use, and prior herpes simplex virus type 2 infection. Cancer Epidemiol Biomarkers Prev 1996; 5:541–548.
22. Smith JS, Herrero R, Bosetti C, et al. Herpes simplex virus-2 as a human papillomavirus cofactor in the etiology of invasive cervical cancer. J Natl Cancer Inst 2002; 94:1604–1613.
23. Ahmed S, Lutalo T, Wawer M, et al. HIV incidence and sexually transmitted disease prevalence associated with condom use: A population study in Rakai, Uganda. AIDS 2001; 15:2171–2179.