Tel.: 410-502-2583, Fax: 410-614-2632,
Cervicovaginal human papillomavirus (HPV)-infection before and after hysterectomy: evidence of different tissue tropism for oncogenic and nononcogenic HPV types in a cohort of HIV-positive and HIV-negative women†
Article first published online: 3 JAN 2012
Copyright © 2011 UICC
International Journal of Cancer
Volume 131, Issue 6, pages 1472–1478, 15 September 2012
How to Cite
D'Souza, G., Burk, R. D., Zhong, Y., Minkoff, H., Stewart Massad, L., Xue, X., Heather Watts, D., Anastos, K., Palefsky, J. M., Levine, A. M., Colie, C., Castle, P. E. and Strickler, H. D. (2012), Cervicovaginal human papillomavirus (HPV)-infection before and after hysterectomy: evidence of different tissue tropism for oncogenic and nononcogenic HPV types in a cohort of HIV-positive and HIV-negative women. Int. J. Cancer, 131: 1472–1478. doi: 10.1002/ijc.27363
HPV DNA testing and statistical analysis in this study was funded through R01-CA085178 (to H.S.). Other data in this article were collected by the Women's Interagency HIV Study (WIHS) Collaborative Study Group with centers (Principal Investigators) at New York City/Bronx Consortium (Kathryn Anastos); Brooklyn, NY (Howard Minkoff); Washington DC Metropolitan Consortium (Mary Young); The Connie Wofsy Study Consortium of Northern California (Ruth Greenblatt); Los Angeles County/Southern California Consortium (Alexandra Levine); Chicago Consortium (Mardge Cohen); Data Coordinating Center (Stephen Gange).
- Issue published online: 20 JUL 2012
- Article first published online: 3 JAN 2012
- Accepted manuscript online: 28 NOV 2011 05:50AM EST
- Manuscript Accepted: 26 SEP 2011
- Manuscript Received: 2 FEB 2011
- National Institute of Allergy and Infectious Diseases. Grant Numbers: UO1-AI-35004, UO1-AI-31834, UO1-AI-34994, UO1-AI-34989, UO1-AI-34993, UO1-AI-42590
- Eunice Kennedy Shriver National Institute of Child Health and Human Development. Grant Number: UO1-HD-32632
- National Cancer Institute
- National Institute on Drug Abuse
- National Institute on Deafness and Other Communication Disorders
- National Center for Research Resources. Grant Number: UL1 RR024131
- viral tropism;
Human papillomavirus (HPV) is detected in nearly all cervical cancers and approximately half of vaginal cancers. However, vaginal cancer is an order of magnitude less common than cervical cancer, not only in the general population but also among women with HIV/AIDS. It is interesting therefore that recent studies found that HPV was common in both normal vaginal and cervical tissue, with higher prevalence of nononcogenic HPV types in the vagina. In our investigation, we prospectively examined HPV infection in 86 HIV-positive and 17 HIV-negative women who underwent hysterectomy during follow-up in a longitudinal cohort. Cervicovaginal lavage specimens were obtained semi-annually and tested for HPV DNA by polymerase chain reaction. To address possible selection biases associated with having a hysterectomy, subjects acted as their own comparison group—before versus after hysterectomy. The average HPV prevalence was higher in HIV-positive than HIV-negative women both before (59% vs. 12%; p < 0.001) and after hysterectomy (56% vs. 6%; p < 0.001). Multivariate random effects models (within-individual comparisons) demonstrated significantly lower HPV prevalence [odds ratio (OR) = 0.71; 95% confidence interval (CI) = 0.59–0.85) after hysterectomy. The association of HPV prevalence with hysterectomy was similar among HIV-positive and HIV-negative women. However, hysterectomy had greater effects on oncogenic (OR = 0.48; 95% CI = 0.35–0.66) than nononcogenic HPV types (OR = 0.89; 95% CI = 0.71–1.11; Pinteraction = 0.002). Overall, we observed greater reductions in oncogenic than nononcogenic HPV prevalence after hysterectomy. If correct, these data could suggest that oncogenic HPV have greater tropism for cervical compared to vaginal epithelium, consistent with the lower incidence of vaginal than cervical cancer.
More than half of vaginal cancers contain oncogenic human papillomavirus (HPV) and, as in cervical cancer, HPV16 is the predominant type present in vaginal tumors that test HPV-positive.1 The incidence of vaginal cancer, however, is much lower than that of cervical cancer (0.4 vs. 8.4 per 100,000 person-years).2 It was therefore a surprise when recent studies in a Costa Rican cohort found that vaginal HPV prevalence was equal to or greater than that in the cervix,3–5 suggesting that the low rates of vaginal cancer could not simply be attributed to a low rate of vaginal HPV infection.
In addition, the type-specific distribution of vaginal HPV in these studies was noteworthy. The initial investigation for example found a higher prevalence of nononcogenic HPV in cervicovaginal specimens among hysterectomized women (reflecting only vaginal tissue) compared to nonhysterectomized women (reflecting both cervical and vaginal tissue), whereas the prevalence of oncogenic HPV did not differ between the two groups.3 Similarly, a second study in the cohort observed a higher prevalence of nononcogenic HPV in vaginal than cervical swabs,4 while a third study found a positive association between the degree of cervical ectopy (an outgrowth of endocervical columnar epithelium onto the ectocervical tissue) and oncogenic HPV infection. Thus, the results from this cohort suggested that oncogenic HPV may have particular tropism for the columnar/metaplastic cells of the cervical squamocolumnar junction than for mature squamous epithelium (present in both the ectocervix and vagina),5 whereas nononcogenic HPV may have greater tropism for vaginal epithelium.
Certain limitations to these data must also be noted, though. Most notably, the specimens tested were from highly selected subgroups of women (e.g., with abnormal pap smears) and each of the studies was cross-sectional.6 There have also been no similar studies in HIV-positive women, a population at high risk of HPV-related tumors.
Despite the high incidence of both vaginal and cervical cancer in women with HIV/AIDS, vaginal cancer rates among HIV-infected women are a small fraction of those observed for the cervix, as in the general population.7 Therefore, studies comparing cervical and vaginal HPV infection in HIV-positive women are warranted. In fact, there are little data at all regarding the prevalence and natural history of vaginal HPV infection in HIV-positive women.
We, therefore, longitudinally examined the prevalent and incident detection of HPV before and then after hysterectomy in HIV-positive and HIV-negative women, using these women as their own comparison group. By comparing these women to themselves (pre- and post-hysterectomy), we hoped to reduce concerns regarding self-selection biases that may be associated with having a hysterectomy.
As described in detail elsewhere,8–10 the Women's Interagency HIV Study (WIHS) is an ongoing cohort that enrolled 2,791 HIV-positive and 975 HIV-negative women from similar clinical and outreach sources during two enrollment periods,1994–1995 and 2001–2002.10 At semi-annual visits, CD4+ T-cell count is determined, a Pap smear and cervicovaginal lavage (CVL) are obtained as well as an interviewer-administered questionnaire. In women who underwent hysterectomy with cervical resection, a vaginal (rather than cervical) Pap smear is obtained and assessed using the 2001 Bethesda System, with colposcopy and biopsy conducted if indicated.11
To minimize the possibility that our findings might be influenced by cases who had a hysterectomy because of HPV-related disease (i.e., eliminating HPV infected cells), we excluded women (N = 9) with evidence of high grade neoplasia at any time pre-hysterectomy. The remaining 103 women (86 HIV-positive and 17 HIV-negative) who had a hysterectomy with cervical resection during follow-up were included in our investigation. These women provided 784 persons-visits pre-hysterectomy and 521 person-visits post-hysterectomy. In a secondary analysis, we also studied 219 women (186 HIV-positive and 33 HIV-negative) who had a hysterectomy before enrollment (HBE) in the WIHS, with a total of 2,766 person-visits of follow-up data (described below).
CVLs were tested for HPV DNA using a MY09/MY11/HMB01 L1 consensus primer polymerase chain reaction (PCR) system. Details of these methods have been previously reported,12 and the results shown to have high reproducibility, specificity and sensitivity (e.g., through comparison to semiquantitative methods such as hybrid capture-2)13–15; demonstrating the clinical relevance of the MY09/MY11/HMB01 PCR assay. Oncogenic types were defined as HPV 16/18/31/33/35/39/45/51/52/56/58/59/68.16, 17 Nononcogenic types were defined as all other HPV types, including 6/11/13/26/32/40/53/54/55/61/62/66/69/70/71/72/ 73/81/82/83/84/85/89, as well as HPV detected only by the generic probe.
In our primary analyses, HPV prevalence before and after hysterectomy was examined using random effects models, and the incident detection of HPV was studied using frailty models. Similar models were used to study time since hysterectomy and detection of both HPV and squamous intraepithelial lesions (SILs); since as mentioned, vaginal cytology was initiated post-hysterectomy. A detailed overview of these statistical methods has been reported18 and were chosen because they allow measurement of the within-individual effects of an exposure variable (e.g., hysterectomy) while adjusting for repeated observations (e.g., multiple HPV types and repeated visits).
Our second analyses examined HPV infection in the above women before their hysterectomies (i.e., censoring at the visit before hysterectomy), and compared these results to those in women who had a HBE. Specifically, HPV prevalence was studied using logistic regression models that incorporated generalized estimating equations (GEE), and incident detection of HPV was studied using Cox models that employed the Wei Lin Weisfeld marginal model approach; methods chosen because they allow measurement of the between-individual effects of hysterectomy, while adjusting for repeated observations.18 The effects of time since hysterectomy on HPV and SIL were studied using similar models. All multivariate models adjusted for age, number of sexual partners in the past 6 months, cigarette smoking and race/ethnicity.
Hysterectomy during follow-up
The dataset included 86 HIV-positive and 17 HIV-negative women who had a hysterectomy during follow-up (HDFs). Table 1 shows selected characteristics of these women at baseline. Compared to women in the WIHS who never had a hysterectomy, at baseline the HDFs were older and more likely to have a prevalent oncogenic HPV infection, and low-grade cervical SIL. However, no differences in race/ethnicity, tobacco use, HIV-serostatus or CD4+ cell count were found between HDFs and women who never had a hysterectomy. At the time of hysterectomy, the median age of the HDFs was 42 years (interquartile range = 38–46).
HIV-positive HDFs were significantly more likely than HIV-negative women to have HPV infection before (average prevalence = 59% vs. 12%; p < 0.001) as well as after hysterectomy (average prevalence = 56% vs. 6%; p < 0.001), and the prevalence of HPV among HIV-seropositive HDFs was positively associated with low CD4+ cell count (Ptrend = 0.03). Although naïve comparisons (using a chi-square test) did not detect an effect of hysterectomy on HPV prevalence (data not shown), multivariate random effects models comparing women to themselves (the within-individual effects) showed a highly significant effect of hysterectomy on HPV prevalence (Table 2). Specifically, in multivariate models that adjusted for other HPV risk factors (see Methods) hysterectomy was associated with lower HPV prevalence [odds ratio (OR) = 0.71; 95% confidence interval (CI) = 0.59–0.85]. Furthermore, while this relationship did not vary by HIV-serostatus (Pinteraction = 0.18), it differed significantly between oncogenic and nononcogenic HPV types (Pinteraction = 0.005); i.e., an effect of hysterectomy was observed for the prevalence of oncogenic (OR = 0.50; 95% CI = 0.37–0.68) but not nononcogenic HPV (OR = 0.88; 95% CI = 0.71–1.10).
A prior study reported that HPV prevalence increased with longer time since hysterectomy,3 and we examined this in our dataset. As shown in Table 2, the prevalence of oncogenic HPV was similar (p = 0.28) ≥3 years and <3 years post-hysterectomy (note: the data were too limited to examine more than two time intervals). In contrast, the prevalence of nononcogenic HPV was higher ≥3 years than <3 years post-hysterectomy (p = 0.003); a result significantly different than for oncogenic HPV (Pinteraction = 0.008).
The incident detection of HPV was also studied (Table 2). In multivariate frailty models, the incident detection of oncogenic HPV ≥3 years post-hysterectomy was low compared to pre-hysterectomy rates [hazard ratio (HR) = 0.34; 95% CI = 0.11–1.02]. These results differed significantly from those for nononcogenic HPV (Pinteraction = 0.003), as the incident detection of nononcogenic HPV ≥3 years post-hysterectomy was nonsignificantly greater than pre-hysterectomy rates (HR = 1.50; 95% CI = 0.87–2.58). After <3 years post-hysterectomy, however, neither oncogenic nor nononcogenic HPV incident detection rates differed significantly from those before hysterectomy.
Finally, we examined vaginal neoplasia in HDFs after hysterectomy, and observed that the prevalence of vaginal SIL was significantly lower (OR = 0.31; 95% CI = 0.15–0.63) ≥3 years post-hysterectomy compared to <3 years post-hysterectomy, controlling for the same cofactors as above. The data were too limited to study incident SIL or to study histologic findings as an endpoint.
Hysterectomy before enrollment
As a second analytic approach, we compared 219 women who had a HBEs to the HDFs before their hysterectomies. Thus, this analysis was conducted amongst women who either had or went on to have a hysterectomy (see Methods). Compared with HDFs the HBEs at baseline were older, more likely to be Black, more likely to be current or former cigarette smokers, less likely to have used oral contraception before baseline and had lower oncogenic HPV prevalence (Table 1).
In multivariate models, HBEs were significantly less likely than the pre-hysterectomy HDFs to have prevalent oncogenic HPV (OR = 0.41; 95% CI = 0.25–0.69), but more likely to have nononcogenic HPV infection (OR = 1.29; 95%CI = 0.97–1.70) (Table 3). The results did not vary significantly by HIV-serostatus (Pinteraction = 0.09). When the analysis was restricted to women who were 35–45 years of age (which we used to make the HBE and HDF age distributions as similar as possible), the results were not qualitatively changed; HBEs were less likely than pre-hysterectomy HDFs to have prevalent oncogenic HPV (OR = 0.76, 95% CI = 0.41–1.39), but more likely to have nononcogenic HPV infection (OR = 1.27; 95% CI = 0.89–1.80, Pinteraction = 0.06). In addition, there was a significant trend of increasing nononcogenic HPV prevalence with greater time since hysterectomy among the HBEs (Ptrend = 0.05), not observed for oncogenic HPV (Table 3).
The incident detection of oncogenic HPV was lower in HBEs (HR = 0.64; 95% CI = 0.39–1.03) than in pre-hysterectomy HDFs, although the association was of borderline significance. The incident detection of nononcogenic HPV did not vary between the two groups (Table 3). The time since hysterectomy had no effect on these findings. Vaginal SIL prevalence also did not vary by time since hysterectomy (Ptrend = 0.58) and, as above, the data were too limited to study incident SIL or histologic findings as an endpoint.
This study prospectively assessed the effects of hysterectomy on rates of HPV infection, using CVL specimens from HIV-positive and HIV-negative women enrolled in a longitudinal cohort. HPV DNA was commonly detected after hysterectomy, suggesting that the vaginal epithelium is a common site of HPV infection. In addition, vaginal HPV prevalence was significantly greater in HIV-positive than in HIV-negative women, and positively associated with the level of HIV-related immunosuppression. There has been little previously reported regarding vaginal HPV infection in HIV-positive women, and these data provide novel evidence that host immunity plays an important role in control of vaginal HPV—findings consistent with reports that rates of vaginal cancer are increased in women with HIV/AIDS.19
Furthermore, hysterectomy was associated with a significant reduction in oncogenic but not nononcogenic HPV prevalence, consistent with recent reports, suggesting that oncogenic HPV may have greater tropism for cervical than vaginal epithelium. The results were similar by HIV status, as well as similar in the two distinct study designs we used: (i) comparing women to themselves, before versus after hysterectomy (our primary analysis); and (ii) comparing women who had a HBE in the cohort to those without current hysterectomy but who later had the procedure. Both approaches helped address concerns regarding potential biases due to differences in the characteristics of women who do and do not undergo hysterectomy.
Interestingly, with increasing time from hysterectomy the prevalence of nononcogenic HPV increased in both analyses, consistent with results of a prior study.4 Oncogenic HPV prevalence, in contrast, remained at a fairly stable but reduced level after hysterectomy. The incident detection of nononcogenic HPV also increased with greater time since hysterectomy in our primary analysis, whereas the incident detection of oncogenic HPV decreased. One possible explanation why oncogenic HPV might have tropism for the cervix is the presence of the transformation zone, the narrow region where columnar cells of the endocervix meet the stratified squamous cells of the ectocervix. Indeed, as mentioned, one of the studies that motivated our investigation reported that greater cervical ectopy was associated with higher oncogenic HPV prevalence.5 However, why nononcogenic HPV infection might increase after hysterectomy is unknown. It has been speculated that vaginal atrophy induced by oophorectomy (removal of the ovaries) may impair the integrity of vaginal tissue,4 increasing the risk of HPV infection with nononcogenic types. If correct, though, oncogenic HPV infection would also be expected to increase over time, even if not to the same extent as with nononcogenic HPV. Unfortunately, accurate data regarding oophorectomy were not obtained in the WIHS and, thus, could not be evaluated in this study. Nonetheless, given that similar results have been obtained in both of the studies in which hysterectomy has been evaluated, further research regarding these relationships is warranted, as it may inform our understanding of HPV viral-host interaction.
Vaginal SIL prevalence decreased after hysterectomy amongst HDFs; results that are in keeping with the decrease we observed in oncogenic HPV infection after hysterectomy, and not with the increase observed in nononcogenic HPV infection. If correct, these data are consistent with a recent WIHS study that showed a stronger relation of vaginal cytologic abnormalities with oncogenic HPV than nononcogenic HPV (Personal communications, L. Stewart Massad).
Several limitations to our study should be considered in interpreting the findings. In particular, the follow-up questionnaire did not ask the reason for hysterectomy in women who had their surgery during the study. Therefore, to minimize the possibility that our primary analyses might have been influenced by cases in which hysterectomy was conducted because of HPV-related disease, we excluded women with evidence of high grade neoplasia at any time before hysterectomy. In addition, the number of women who underwent hysterectomy during WIHS follow-up was relatively small, leading to a modest sample size in this study.
It must be noted that hysterectomy should not be undertaken as a cancer prevention measure in the absence of other indications in women with HIV. Data from our study and others have shown that with regular screening and treatment for cervical neoplasia, the risk of cervical cancer, as well as the risk of vaginal cancer, is low in HIV-positive women.20–22 On the other hand, while HIV-negative women can stop Pap testing after hysterectomy in the absence of high-grade cervical neoplasia or cancer, it is unclear whether the risk of vaginal cancer is sufficiently low to allow HIV-positive women to stop Pap testing after hysterectomy.23
In summary, this study had three major findings, namely, that HPV infection of the vagina is common, that the rate of vaginal HPV infection is increased in immunosuppressed HIV-positive women and, more speculatively, that oncogenic HPV may preferentially infect cervical compared to vaginal epithelium in both HIV-positive and HIV-negative women. Further research to better understand the biologic factors that underlie differences in HPV viral tropism may help to identify targets useful in chemoprevention.