Human papillomavirus (HPV) is widely considered to be the causative agent of cervical cancer,1 a disease diagnosed in nearly 500,000 women annually worldwide.2 Although vaccines targeting the most virulent oncogenic HPV types hold great promise,3 current vaccines are expensive. Sustainable and affordable strategies to control spread of HPV, especially in low-resource settings, are of major public health importance.
Multiple observational studies demonstrate consistent associations between barrier contraceptive use (both condoms and diaphragms) and decreased risk of HPV infection4,5 and cervical neoplasia.6–15 The U.S. Preventive Services Task Force summarized evidence on primary prevention of cervical cancer and reported that diaphragm use alone was associated with a 30–80% decreased likelihood of cervical cancer.16 Although a randomized trial of condoms demonstrated efficacy for both regression of precancerous cervical lesions and HPV infections,17 no randomized trials have been performed to determine the effect of diaphragm use on HPV infection or cervical neoplasia. We estimated the effect of provision of a latex diaphragm, lubricant gel, and male condoms (intervention) compared with condoms alone (control) on incidence and clearance of HPV infection in the setting of a randomized trial.
MATERIALS AND METHODS
Participants were sexually active HIV-negative women enrolled in a multicentered (Zimbabwe and South Africa), open-label, randomized trial estimating 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 (MIRA) study (ClinicalTrials.gov number NCT00211459), are presented elsewhere.18
The study reported here was conducted only at the Zimbabwe site at two clinics: Chitungwiza, a periurban municipality near Harare; and Epworth, a poorer and less developed suburb of Harare. Women were recruited from September 2003 to September 2005. Beginning February 2004, enrollment in the HPV study was offered to main trial enrollees; 2,040 of 2,089 (97.6%) women accepted and gave written informed consent. As per the main study protocol, all women were counseled about and provided male condoms at trial entry and at each visit. Counseling included information on the importance of condom use at all sex acts in the prevention of HIV.
Randomly permuted blocks of sizes 8, 10, and 12 were used for the randomization scheme. At the point of randomization, a sealed, opaque envelope containing the randomization assignment was retrieved by study staff and opened by the participant. Those assigned to the intervention group received a clinician-fitted latex diaphragm (Ortho-McNeil Pharmaceutical, Raritan, NJ), a supply of lubricant gel (Replens, Lil’ Drug Store Products, Cedar Rapids, IA), and a supply of male condoms; the control group received condoms alone.
The study protocol was reviewed and approved by the University of California, San Francisco Committee on Human Research and by the local ethics review committee, the Medical Research Council of Zimbabwe, and the Medicines Control Authority of Zimbabwe.
At the baseline visit, all women underwent a speculum examination. Clinicians collected specimens for cervical cytology using a spatula and an endocervical brush. For HPV testing, a polyester swab was used to swab the cervix and endocervix under direct visualization; the swab was placed in specimen transport medium (Digene Diagnostics, Silver Spring, MD) and stored initially at 4°C for up to 4 days and subsequently at –20°C. Self-sampling was performed at 3-month intervals at the regularly scheduled clinic visits. Participants were instructed to place the swab into the vagina as far as it would comfortably go and twist the swab around its axis; the procedure was repeated with a second swab. Participants placed both swabs in transport medium. At the 12-month visit, clinicians performed an additional speculum examination to collect a cervical sample for HPV testing after the self-collected swab was obtained. At study exit, only a clinician-collected sample was obtained. At each visit, information about condom use and adherence to study products was determined by structured interview.
Samples were shipped monthly to University of California, San Francisco (Palefsky laboratory) for analysis using L1 consensus polymerase chain reaction primers and primers for amplification of the human β-globin gene. Thawed tubes were heated to 56°C for 1 hour to inactivate viruses and then digested overnight with proteinase K19; an aliquot was precipitated and concentrated. After 30 amplification cycles, 6 microliters of amplification mixture was applied to a nylon membrane and probed with a biotin-labeled HPV L1 consensus probe mixture. A separate membrane was probed with a biotin-labeled probe to the human β-globin gene. Specimens positive with the HPV consensus probe mixture were then assayed for the following 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, a variant of 82, 83, and 84; the following 10 HPV types were included together in a probe mixture: 2, 13, 34, 42, 57, 62, 64, 67, 72, and 82 (known as “mixed” types). Specimens positive for the generic probe, but negative for the 39 specific types were considered to have an “untyped” sample. We defined oncogenic HPV as types 16, 18, 31, 33, 35, 39, 45, 51, 52, 56, 58, 59, and 68.
We anticipated that 2,200 women would enroll in Zimbabwean sites and that 10% would be lost to follow-up. Assuming an HPV prevalence ranging from 20% to 40% and an annual HPV incidence of 8%, we estimated that we would have 80% power to detect relative hazards of at least 0.63 for the high prevalence estimate and 0.74 for the low estimate, (two-tailed alpha=0.05). We considered reductions of this magnitude to be clinically important. The primary outcomes were based on results of clinician-collected HPV testing (up to three tests per participant). In an intention-to-treat analysis, we evaluated the effect of diaphragm/gel provision on prevalence of HPV at the time of the first postenrollment clinician-collected test. We focused on detection of any HPV type, any oncogenic HPV type, and the most common oncogenic HPV types in Zimbabwe (types 16, 18, 33, and 58, as determined by prevalence at baseline).
We defined HPV incidence in two ways: 1) detection of any “new” HPV type(s) (ie, one not detected at baseline) among all participants, and 2) detection of any HPV among women with no HPV detected at baseline. We estimated HPV incidence proportion at two predefined time points: 12 months and study exit. We estimated incidence of any oncogenic HPV and incidence of HPV types 16, 18, 33, and 58 using these two definitions. We defined a “12-month” test as one performed 11 to 16 months after the baseline HPV test; tests performed after 16 months (but within 2 weeks of study closure) were defined as “exit” tests.
We defined HPV clearance as nondetection of one or more HPV type(s) among women with any HPV detected at baseline. Samples that were untyped at baseline were not considered cleared if a specific HPV type was detected at follow-up. We estimated overall clearance of all oncogenic HPV types and type-specific clearance of HPV types 16, 18, 33, and 58, defined as nondetection of these types among women with one or more of these types detected at baseline. We estimated HPV clearance proportion at 12 months and study exit.
Using results from both self- and clinician-collected tests (up to nine tests per participant), we also used the Kaplan-Meier method to estimate the cumulative probability of HPV occurrence across the entire study period for each randomized group. We used a similar approach to estimate type-specific HPV clearance, defined as two subsequent consecutive tests negative for the HPV type detected at baseline and no future tests positive for that type.
We performed a “per-protocol” analysis by restricting our main analyses to women reporting diaphragm/gel use at 100% of prior sex acts. In sensitivity analyses, we expanded our definition of oncogenic HPV to include five additional HPV types (26, 53, 66, 73, and 82).20 In intention-to-treat analyses, we excluded HPV tests with no detectable β-globin (n=99, or 4.9% of clinician-collected tests) and those performed more than 14 days after study exit (n=8); we also excluded women with no postenrollment HPV tests due either to lost to follow-up or study withdrawal (n=107). In all other analyses, we additionally excluded women with a missing baseline HPV test (n=5) and those with untyped HPV at both baseline and follow-up (n=16). We calculated relative risks and 95% confidence intervals for outcomes at discrete preestablished time points (time of first postenrollment test, at 12 months, and at study exit). Between-group differences in cumulative incidence and clearance probabilities were assessed using the log rank test.
The Bill and Melinda Gates Foundation funded and approved the HPV study but had no other role in data collection, analysis, data interpretation or reporting of results. All authors had access to the data and agreed with the results as submitted for publication.
Figure 1 describes the participants’ progression through the phases of the trial. Characteristics of study participants were similar between groups at baseline (Table 1); the baseline prevalence of HPV type 18 was somewhat higher in the intervention group (P=.05). In an intention-to-treat analysis, HPV prevalence at the time of the first postenrollment clinician-collected test was the same in the two groups (Table 2). In the intervention group, prevalence of HPV types 18 and 58 was higher at the time of the first postenrollment HPV test.
Table 3 shows HPV incidence at 12 months by randomized group; no differences were demonstrable for any outcome. Results were similar at study exit (data not shown). Cumulative HPV incidence rates using results of both self- and clinician-collected HPV tests among women with no HPV detected at baseline were similar in the two study arms (Fig. 2A, P=.71). Table 4 shows HPV clearance at 12 months and exit by randomized group. For most outcomes, no differences were noted between randomized groups. At 12 months, clearance of oncogenic HPV (to “no HPV detected”) and clearance of HPV type 58 were lower in the intervention group; neither difference was statistically significant at exit. Clearance of HPV type 18 was lower at exit in the intervention group.
Using results of both self- and clinician-collected HPV tests, Kaplan-Meier estimates of cumulative HPV clearance probabilities among women with any HPV detected at baseline (to “no HPV detected”) and women with any oncogenic HPV detected at baseline (to “no HPV detected”) were not significantly different between randomized groups (any HPV, Fig. 2B, P=.16; any oncogenic HPV, P=.72). Estimated cumulative clearance probabilities of HPV type 18 (P=.11) and HPV type 58 (P=.17) were also similar between randomized groups.
Women in the intervention group were more likely to report infrequent condom use, defined as condom use at less than one third of last sex acts (23% compared with 13%, P<.001). At 12 months, about one half of women in the intervention group (50.5%) reported diaphragm/gel use at 100% of prior sex acts; at study exit, this proportion was 34.5%. In “per-protocol” analyses, women reporting diaphragm/gel use at 100% of prior sex acts had a lower likelihood of having one or more new HPV types detected at 12 months (17.8% compared with 23.9%, RR 0.75, 95% CI 0.58–0.96) and exit (21.4% compared with 27.9%, RR 0.77, 95% CI 0.59–0.99) compared with women in the control group. Women in the intervention group with no HPV at baseline had a lower likelihood of having any HPV detected at 12 months (16.2% compared with 22.3%, RR 0.73, 95% CI 0.54–0.98); this difference was not statistically significant at exit (20.0% compared with 25.9%, RR 0.77, 95% CI 0.57–1.05). Human papillomavirus clearance was not affected by consistency of diaphragm/gel use. Expanding our definition of oncogenic HPV to include five additional types yielded similar results.
Diaphragm plus lubricant gel provision did not affect HPV incidence or clearance among women receiving risk reduction counseling and condoms in an HIV prevention program. Although our findings do not support the protective effects of diaphragm use on HPV infection reported in observational studies, the specific context in which this study was performed may limit its generalizability to other settings. Although restricting analyses to women with high adherence to diaphragm/gel use showed some protection from incident HPV infections, these data must be interpreted with caution due to the potential for substantial confounding with subgroup analyses, especially because a large proportion of women in the intervention group were excluded from these analyses.
Decreased clearance of HPV types 18 and 58 was noted at the two predefined time points among women in the intervention group. These differences were not statistically significant over the course of the entire trial; a higher proportion of women with HPV type 18 at baseline in the intervention group may have accounted for the HPV type 18–related observation. Of note, type-specific analyses were limited by small numbers of outcomes and subsequent low statistical power. Given that women in the intervention group were more likely to report infrequent condom use, these observed differences may have been due to less condom use among women in the intervention arm or due to actual use of the diaphragm/gel.
Our study has several limitations. We could only estimate HPV incidence and clearance within each arm because we had a maximum of three time points when clinician-collected swabs were obtained; some HPV tests may have been either falsely positive or falsely negative such that true incidence and clearance may have been either underestimated or overestimated. Errors due to false-positive and false-negative testing, however, would be expected to occur equally between the two groups because of randomization; the relative comparisons should reflect valid differences between groups. Our intervention was diaphragm/gel provision, and we could not fully ascertain how often and how correctly participants used the study products because we relied only on self-report; poor adherence to diaphragm/gel use may have accounted for the null effect observed. Because the trial could not be blinded, assignment to the intervention group seems to have had an effect on condom use, and may have had an effect on reporting of study product use. Numerous other explanations based on equally plausible differences due to knowledge of the group assignment may also explain our findings. Another plausible explanation for our observed lack of effect may be a beneficial effect of the intervention on HPV that was offset by a detrimental effect due to concurrent decrease in the use of condoms among women in the intervention group.
Study strengths include its randomized design, large size, and small numbers of participants lost to follow-up. We were able to exclude potential effects of the lubricant gel on HPV detection based on studies we conducted before completion of the trial.21
Because of the ethical imperative to offer condoms to both arms in this trial of HIV seroconversion, we could only estimate the marginal benefit of diaphragm/gel provision beyond that provided by condoms alone. Given evidence demonstrating benefit of condoms for clearance of HPV infection,17 additional benefit contributed by diaphragm/gel may not have been measurable. Our findings in the per-protocol analysis and the importance of female-controlled methods of prevention of sexually transmissible infections, however, indicate that further study of diaphragms for the prevention of HPV and HPV-related diseases may be warranted.
1. Schiffman M, Castle PE, Jeronimo J, Rodriguez AC, Wacholder S. Human papillomavirus and cervical cancer. Lancet 2007;370:890–907.
2. Castellsague X, de Sanjose S, Aguado T, Louie KS, Bruni L, Munoz J, et al. HPV and cervical cancer in the world 2007 report. Vaccine 2007;25 suppl:C1–230.
3. Sawaya GF, Smith-McCune K. HPV vaccination–more answers, more questions. N Engl J Med 2007;356:1991–3.
4. Winer RL, Hughes JP, Feng Q, O’Reilly S, Kiviat NB, Holmes KK, et al. Condom use and the risk of genital human papillomavirus infection in young women. N Engl J Med 2006;354:2645–54.
5. Kjaer SK, van den Brule AJ, Bock JE, Poll PA, Engholm G, Sherman ME, et al. Determinants for genital human papillomavirus (HPV) infection in 1000 randomly chosen young Danish women with normal Pap smear: are there different risk profiles for oncogenic and nononcogenic HPV types? Cancer Epidemiol Biomarkers Prev 1997;6:799–805.
6. Kjaer SK. Risk factors for cervical neoplasia in Denmark. APMIS Suppl 1998;80:1–41.
7. Kjaer SK, de Villiers EM, Caglayan H, Svare E, Haugaard BJ, Engholm G, et al. Human papillomavirus, herpes simplex virus and other potential risk factors for cervical cancer in a high-risk area (Greenland) and a low-risk area (Denmark)—a second look. Br J Cancer 1993;67:830–7.
8. Coker AL, Sanders LC, Bond SM, Gerasimova T, Pirisi L. Hormonal and barrier methods of contraception, oncogenic human papillomaviruses, and cervical squamous intraepithelial lesion development. J Womens Health Gend Based Med 2001;10:441–9.
9. Parazzini F, Negri E, La Vecchia C, Fedele L. Barrier methods of contraception and the risk of cervical neoplasia. Contraception 1989;40:519–30.
10. Peters RK, Thomas D, Hagan DG, Mack TM, Henderson BE. Risk factors for invasive cervical cancer among Latinas and non-Latinas in Los Angeles County. J Natl Cancer Inst 1986;77:1063–77.
11. Wright NH, Vessey MP, Kenward B, McPherson K, Doll R. Neoplasia and dysplasia of the cervix uteri and contraception: a possible protective effect of the diaphragm. Br J Cancer 1978;38:273–9.
12. Celentano DD, Klassen AC, Weisman CS, Rosenshein NB. The role of contraceptive use in cervical cancer: the Maryland Cervical Cancer Case–Control Study. Am J Epidemiol 1987;126:592–604.
13. Jones CJ, Brinton LA, Hamman RF, Stolley PD, Lehman HF, Levine RS, et al. Risk factors for in situ cervical cancer: results from a case-control study. Cancer Res 1990;50:3657–62.
14. Fasal E, Simmons ME, Kampert JB. Factors associated with high and low risk of cervical neoplasia. J Natl Cancer Inst 1981;66:631–6.
15. Becker TM, Wheeler CM, McGough NS, Stidley CA, Parmenter CA, Dorin MH, et al. Contraceptive and reproductive risks for cervical dysplasia in southwestern Hispanic and non-Hispanic white women. Int J Epidemiol 1994;23:913–22.
16. U.S. Preventive Services Task Force. Guide to clinical preventive services: an assessment of the effectiveness of 169 interventions: report of the U.S. Preventive Services Task Force. Baltimore (MD): Williams & Wilkins;1996.
17. Hogewoning CJ, Bleeker MC, van den Brule AJ, Voorhorst FJ, Snijders PJ, Berkhof J, et al. Condom use promotes regression of cervical intraepithelial neoplasia and clearance of human papillomavirus: a randomized clinical trial. Int J Cancer 2003;107:811–6.
18. Padian NS, van der Straten A, Ramjee G, Chipato T, de Bruyn G, Blanchard K, et al. Diaphragm and lubricant gel for prevention of HIV acquisition in southern African women: a randomised controlled trial. Lancet 2007;370:251–61.
19. Palefsky JM, Holly EA, Ralston ML, Jay N. Prevalence and risk factors for human papillomavirus infection of the anal canal in human immunodeficiency virus (HIV)–positive and HIV-negative homosexual men. J Infect Dis 1998;177:361–7.
20. Munoz N, Bosch FX, de Sanjose S, Herrero R, Castellsague X, Shah KV, et al. Epidemiologic classification of human papillomavirus types associated with cervical cancer. N Engl J Med 2003;348:518–27.
21. Smith-McCune KK, Tuveson JL, Rubin MM, DaCosta MM, Darragh TM, Shiboski SC, et al. Effect of Replens gel used with a diaphragm on tests for human papillomavirus and other lower genital tract infections. J Low Genit Tract Dis 2006;10:213–8.