Infectious Causes of Cancer
Time to clearance of human papillomavirus infection by type and human immunodeficiency virus serostatus
Article first published online: 27 APR 2006
Copyright © 2006 Wiley-Liss, Inc.
International Journal of Cancer
Volume 119, Issue 7, pages 1623–1629, 1 October 2006
How to Cite
Koshiol, J. E., Schroeder, J. C., Jamieson, D. J., Marshall, S. W., Duerr, A., Heilig, C. M., Shah, K. V., Klein, R. S., Cu-Uvin, S., Schuman, P., Celentano, D. and Smith, J. S. (2006), Time to clearance of human papillomavirus infection by type and human immunodeficiency virus serostatus. Int. J. Cancer, 119: 1623–1629. doi: 10.1002/ijc.22015
- Issue published online: 18 JUL 2006
- Article first published online: 27 APR 2006
- Manuscript Accepted: 16 FEB 2006
- Manuscript Received: 29 JAN 2006
- National Cancer Institute National Research Service Award. Grant Number: CA09330-22
- Centers for Disease Control and Prevention. Grant Numbers: U64/CCU106795, U64/CCU206798, U64/CCU306802, U64/CCU506831
- Center for AIDS Research. Grant Number: 5 P30 AI050410-07 NIAID
- HPV type;
- time to clearance
Persistent infection with high-risk human papillomavirus (HPV) is central to cervical carcinogenesis. Certain high-risk types, such as HPV16, may be more persistent than other HPV types, and type-specific HPV persistence may differ by HIV serostatus. This study evaluated the association between HPV type and clearance of HPV infections in 522 HIV-seropositive and 279 HIV-seronegative participants in the HIV Epidemiology Research Study (HERS, United States, 1993–2000). Type-specific HPV infections were detected using MY09/MY11/HMB01-based PCR and 26 HPV type-specific probes. The estimated duration of type-specific infections was measured from the first HPV-positive visit to the first of two consecutive negative visits. Hazard ratios (HRs) and 95% confidence intervals (CIs) for HPV clearance were calculated using Cox models adjusted for study site and risk behavior (sexual or injection drugs). A total of 1,800 HPV infections were detected in 801 women with 4.4 years median follow-up. HRs for clearance of HPV16 and related types versus low-risk HPV types were 0.79 (95% CI: 0.64–0.97) in HIV-positive women and 0.86 (95% CI: 0.59–1.27) in HIV-negative women. HRs for HPV18 versus low-risk types were 0.80 (95% CI: 0.56–1.16) and 0.57 (95% CI: 0.22–1.45) for HIV-positive and -negative women, respectively. HPV types within the high-risk category had low estimated clearance rates relative to low-risk types, but HRs were not substantially modified by HIV serostatus. © 2006 Wiley-Liss, Inc.
International estimates indicate that over 273,000 women died in 2002 due to invasive cervical cancer.1 Although human papillomavirus (HPV) infection is the central causative agent in cervical carcinogenesis,2 HPV infection alone is not a sufficient cause of cervical cancer. Most HPV infections are transient,3, 4 and women with persistent HPV infections are at increased risk of cervical lesions and cancer.5, 6, 7, 8 Thus, the identification of specific factors that increase the risk of persistent HPV infection may help identify women at an increased risk of cervical cancer.
HPV types that are most prevalent in invasive cervical cancer are classified as “high-risk.”9 Persistence may vary by HPV type, but previous PCR-based studies examining the duration of high-risk versus low-risk HPV infections have been inconsistent, with some reporting positive associations between high-risk HPV types and persistence,3, 10, 11 while others have reported no association.12 Part of the discrepancy may have been due to differences in the specific HPV types included or low statistical power, or alternatively, due to differences in the natural history of specific HPV types classified as “high-risk,” since individual high-risk types may contribute differently to cervical carcinogenesis. For example, both HPV16 and 18, the two most common cancer-associated HPV types worldwide,13 are strongly associated with invasive cervical cancer,9 but these types may differ with regard to infection duration and cervical carcinogenesis. In vitro, HPV18 has been shown to transform human keratinocytes more efficiently than HPV16, which suggests a potential for a faster progression from HPV18 infection to invasive cervical cancer.14, 15
Reports of the relative duration of HPV16 and HPV18 infections have been inconsistent.3, 11, 16, 17, 18, 19, 20 A study of women with normal cytology in the US reported that HPV16 infections persisted longer than other high-risk HPV infections,16 and HPV16 infections had a longer median duration than HPV18 infections in 2 studies of young women in the United Kingdom and Canada.19, 20 In addition, HPV16 infections cleared more slowly than low-risk HPV infections in a cohort of Colombian women (hazard ratio [HR]: 0.47; 95% confidence interval [CI]: 0.32–0.72), while HPV18 and phylogenetically related types cleared at approximately the same rate as low-risk types (HR: 0.98; 95% CI: 0.70–1.39).21 In contrast, studies of low-income women in Brazil,11 young women in the US,17 and HIV-seropositive and HIV-seronegative women in the US18 have reported that HPV16 persistence was comparable to other high-risk HPV types, and the median duration of HPV16 and HPV18 infections was similar in a study of young US women.3
Infection with HIV has been positively associated with a higher risk of persistent HPV infection12 and cervical dysplasia.22, 23 HIV infection (or immune suppression) may also modify associations between HPV type and the duration of HPV infection. Previously, the persistence of “high-risk” and “low-risk” HPV infections was examined among HIV-seropositive and HIV-seronegative women in the multicenter HIV Epidemiology Research Study (HERS) cohort.12 This report describes a more detailed analysis of the time required to clear infections with specific HPV types (HPV16 and HPV18) and subgroups of high-risk types among HIV-seropositive and -seronegative women in the HERS cohort.
Material and methods
Data were evaluated from the HERS cohort of 871 HIV-seropositive and 439 HIV-seronegative women enrolled at 4 sites (Baltimore, MD; Bronx, NY; Detroit, MI; and Providence, RI), as previously described.24 Women were eligible for HERS if they had HIV without an AIDS-defining condition (1987 Centers for Disease Control and Prevention case definition)25 or if they were at risk of HIV infection due to injection drug use or high-risk sexual behavior (5 or more sexual partners within 5 years; exchanging sex for money or drugs; or having had sex with a male injection drug user, or a known or suspected HIV-positive male).
By study design, 50% of the women were enrolled based on injection drug use risk behavior and 50% because of high-risk sexual behavior. All sites maintained Public Health Service Certificates of Confidentiality and received approval from local institutional review boards.
HERS study visits were scheduled at 6-month intervals, with a maximum of 15 visits. Each visit included a gynecological exam, cervicovaginal lavage to collect samples for detecting HPV DNA, and Papanicolaou test screening (which was annual after visit 10). Women were eligible for the current study if they had 2 or more study visits, baseline data on HPV DNA, a cervix, no cervical treatment in the 6 months before enrollment, and no low- or high-grade squamous intraepithelial lesions or invasive cancer at baseline. Women with HIV seroconversion (n = 11) or pregnancy (n = 88) during follow-up were excluded. Person time was censored at hysterectomy (n = 18), cervical treatment (n = 33), or after the tenth visit.
HPV infections, rather than women, were treated as the unit of analysis because women could be simultaneously infected with multiple HPV types and be infected with the same HPV type more than once. PCR with MY09/MY11/HMB0l primers was used for HPV detection, and HPV type was determined by hybridization with individual probes for 26 types, as previously described.10 Samples that were positive with general probes for HPV DNA but negative for all type-specific probes were classified as untyped. DNA quality was evaluated by amplifying a segment of the human β-globin gene; samples negative for β-globin were categorized as missing HPV DNA data (n = 13).
The duration of each type-specific infection was estimated by setting the start date as the first visit positive for a specific HPV type and the end date as the date of the first of 2 consecutive visits negative for the same HPV type. Infections were classified as “prevalent” if they were detected at the baseline visit and as “incident” infections if they were first detected after the baseline visit. HPV infections followed by a single HPV-negative visit and no further follow-up were censored at the HPV-negative visit. When follow-up ended after an HPV positive visit, 6 months were added to the duration of the HPV infection to ensure comparability of assumptions about clearance for censored and noncensored observations. Neither the six-month addition nor mid-interval assumptions about the duration of HPV infections appreciably changed HPV clearance results (data not shown). Type-specific HPV infections spanning a single missing visit were assumed to persist over that period. Infections spanning 2 or more consecutive missing visits were excluded (n = 55).
HPV infection over time was examined using Kaplan–Meier curves and the log-rank test of homogeneity, and Cox proportional hazards models were used to evaluate associations between HPV types and time to clearance of HPV infections. HPV type categories included the following: (a) low-risk HPV, (b) high-risk HPV, (c) HPV16, (d) HPV16-related types (i.e. types more closely related phylogenetically to HPV16 than to other types), (e) HPV16 and HPV16-related types combined, (f) HPV18, (g) HPV18-related types, (h) HPV18 and HPV18-related types combined, and (i) “other high-risk” HPV types (Table I). High- and low-risk HPV categories were derived from an international case control analysis9 and a recent study that suggested HPV53 should be classified as a low-risk type.26
|No. of HPV infections (% of HPV positives)||Duration (years)||No. of HPV infections (% of HPV positives)||Duration (years)|
|HPV16 and related|
|16||57 (3.9)||0.7||0.3–5.1||23 (6.8)||1.0||0.5–4.9|
|31||47 (3.2)||1.1||0.4–4.9||18 (5.3)||1.0||0.4–2.9|
|33||48 (3.3)||1.0||0.4–4.9||5 (1.5)||0.6||0.5–1.5|
|35||37 (2.5)||1.0||0.2–3.9||7 (2.1)||0.7||0.5–1.8|
|52||55 (3.8)||1.0||0.2–4.5||13 (3.8)||0.7||0.4–1.6|
|58||60 (4.1)||1.3||0.4–5.0||14 (4.1)||0.6||0.4–5.0|
|Any||304 (20.8)||1.0||0.2–5.1||80 (23.5)||0.8||0.4–5.0|
|HPV18 and related|
|18||62 (4.3)||1.0||0.4–4.0||11 (3.2)||0.8||0.5–2.0|
|39||11 (0.8)||1.0||0.5–5.0||3 (0.9)||0.5||0.2–1.1|
|45||42 (2.9)||1.2||0.4–4.9||7 (2.1)||0.7||0.5–4.1|
|59||13 (0.9)||0.7||0.5–3.6||5 (1.5)||1.1||0.5–1.5|
|68||46 (3.2)||0.7||0.4–4.5||6 (1.8)||0.5||0.5–3.2|
|Any||174 (11.9)||1.0||0.4–5.0||32 (9.4)||0.7||0.2–3.2|
|26||13 (0.9)||0.5||0.4–1.0||0 (0.0)||–||–|
|51||69 (4.7)||0.6||0.3–4.1||5 (1.5)||0.5||0.5–4.1|
|56||43 (3.0)||0.9||0.3–5.1||7 (2.1)||0.5||0.4–1.0|
|66||41 (2.8)||0.9||0.2–5.0||8 (2.4)||0.6||0.4–1.4|
|73||32 (2.2)||0.5||0.3–3.5||3 (0.9)||0.6||0.5–1.4|
|82||15 (1.0)||0.5||0.4–3.5||2 (0.6)||0.8||0.5–1.1|
|Any||213 (14.6)||0.6||0.2–5.1||25 (7.4)||0.5||0.4–4.1|
|06||48 (3.3)||1.0||0.1–5.1||4 (1.2)||1.3||0.9–1.7|
|11||15 (1.0)||0.9||0.4–4.5||2 (0.6)||1.0||0.5–1.4|
|40||13 (0.9)||0.5||0.3–1.5||1 (0.3)||0.5||0.5–0.5|
|42||0 (0.0)||–||–||0 (0.0)||–||–|
|53||123 (8.4)||0.9||0.2–4.9||21 (6.2)||0.5||0.3–5.0|
|54||66 (4.5)||0.6||0.3–5.1||16 (4.7)||0.6||0.3–3.0|
|55||35 (2.4)||0.5||0.3–3.0||2 (0.6)||0.5||0.5–0.5|
|83||82 (5.6)||1.0||0.4–4.4||24 (7.1)||0.5||0.4–4.0|
|84||73 (5.0)||0.7||0.2–4.9||12 (3.5)||0.5||0.3–1.9|
|Any||455 (31.2)||0.8||0.1–5.1||82 (24.1)||0.5||0.3–5.0|
|Untyped||314 (21.5)||0.6||0.2–5.1||121 (35.6)||0.5||0.3–4.5|
|Total||1460 (100.0)||0.8||0.1–5.1||340 (100.0)||0.6||0.2–5.0|
Our main analyses compared estimated time to clearance for infections in individual high-risk HPV categories relative to infections in the low-risk category. HRs less than 1.0 are therefore consistent with lower clearance rates (more persistent infections) for the high-risk HPV type or category being evaluated relative to the low-risk (referent) group. Additional analyses compared clearance of HPV16 and related types to all other high-risk types and to all other HPV types, HPV18 and related types to all other high-risk types and to all other HPV types, and HPV16 and related types to HPV18 and related types. Because of their unknown oncogenic status, untyped infections were included only in analyses that used all other HPV types as the referent. Results are presented separately for HIV-seropositive women and HIV-seronegative women. Unless otherwise stated, results are presented for combined “prevalent” (present at baseline) and “incident” (not present at baseline) HPV infections; however, results of models restricted to “incident” infections are presented for selected analyses.
All models adjusted for HERS study site and baseline risk classification (sexual behavior or injection drug use). Potential confounders included baseline age, ethnicity, education, ever smoking, number of recent (last 6 months) male sexual partners, recent condom use, frequency of vaginal sex, recent use of hormonal contraceptives, pregnancy history, bacterial vaginosis, Trichomonas vaginalis infection, hepatitis C serum antibodies, and herpes simplex virus types 1 and 2 serum antibodies. CD4 count, HIV viral load, and antiretroviral therapy at the start of HPV infection were also evaluated as potential confounders in analyses restricted to HIV-positive women. Potential confounders would have been retained if removing them from a model changed HRs for HPV16, HPV16-related, HPV18, HPV18-related, or “other high-risk” types versus low-risk types by 15% or more;27 however, none of the potential confounders met this criterion. Adjusting for multiple HPV infections (overall or high-risk) also did not affect HRs.
Departures from multiplicative effects (modification of the HR) for (a) HIV serostatus and any high-risk HPV type, (b) HIV serostatus and HPV16 and related infections, and (c) HIV serostatus and HPV18 and related infections were evaluated using a likelihood ratio test for interaction terms.28 Because one woman could contribute multiple HPV infections to the analysis, estimates of variance were derived from a robust variance estimator that accounted for within-subject correlation.29, 30 Analyses including continuous and categorical time interaction terms suggested no violation of the proportional hazards assumption.31 Our analysis and interpretation of results focused on patterns of association defined by the magnitude and precision of the HRs rather than solely on statistical significance.
The final study cohort consisted of 801 women (522 HIV-seropositive, 279 HIV-seronegative) with a total of 6,061 visits and a mean age of 35 years. Median follow-up was 4.4 years for both HIV-positive and HIV-negative women, and 88% of study participants had 4 or more visits. Overall, 619 women were HPV positive at least once, and 1,800 HPV infections were detected with an observed overall median duration of 0.7 years. Fifty-five percent of HIV-positive women (n = 287) and 26% of HIV-negative women (n = 73) had infections detected at the baseline visit (i.e., “prevalent” infections).
Among HIV-positive women, HPV53 was the most common individual type (123 infections, 8.4%) and HPV51 was the most common high-risk type (69 infections, 4.7%) (Table I). Among HIV-negative women, HPV83 was the most common individual type (24 infections, 7.1%) and HPV16 was the most common high-risk type (23 infections, 6.8%). Low-risk HPV infections accounted for the highest proportion of typed infections in both HIV-positive women (455 of 1,460 total infections, 31.2%) and HIV-negative women (82 of 340 infections, 24.1%).
The estimated median duration of all HPV infections combined was 0.8 years in HIV-positive women and 0.6 years in HIV-negative women (Table I) (2-sample median test p-value = 0.008). The estimated median duration of infections was longer for the majority of individual HPV types among HIV-positive versus HIV-negative women, with the notable exception of HPV16 (median duration of 0.7 and 1.0 years (2-sample median test p-value = 0.81) among HIV-positive and HIV-negative women, respectively). HPV16 and related type infections and HPV18 and related type infections had longer median durations than other HPV type categories among both HIV-positive women (median duration of 1.0 years for both categories) and HIV-negative women (median duration of 0.8 and 0.7 years, respectively.) The median duration of high-risk infections was the same for women aged 35 or less and women over age 35 (0.9 years), while the median duration of low-risk infections was 0.9 and 0.6 years in women 35 or less and over 35, respectively. Kaplan–Meier curves and the log-rank test for the proportion of HPV infections cleared over time in HIV-positive women indicated a statistically significant difference between HPV risk groups (p = 0.006) (Fig. 1a). The time to 50% clearance was longest for HPV16 and related types (1.9 years), followed by HPV18 and related types (1.5 years), low-risk HPV types (1.4 years), and “other high-risk” HPV types (1.0 years). Among HIV-negative women, 50% of HPV infections cleared in similar time periods for HPV16 and related types (0.99 years), HPV18 and related types (0.97 years), and low-risk HPV types (0.94 years), but 50% clearance occurred more quickly for other “high-risk” HPV types (0.57 years) (Fig. 1b). However, the number of clearance events in HIV-negative women was small, leading to substantial imprecision in the Kaplan–Meier curves.
Estimated HRs for the clearance of prevalent and incident HPV infections (combined) suggest a similar rate of clearance for all high-risk HPV types relative to low-risk types among HIV-positive and -negative women (Table II). However, results for HIV-positive women support a somewhat reduced rate of clearance for HPV16 and related types (HR: 0.79, 95% CI: 0.64–0.97) and slight reduction in clearance of HPV18 and related types (HR: 0.86, 95% CI: 0.67–1.11) relative to low-risk HPV types, while results for HIV-negative women suggest a slightly reduced rate of clearance for HPV-16 and related type infections relative to low-risk type infections only (HR: 0.86, 95% CI: 0.59–1.27). Although the estimate was imprecise, HPV18 infections appeared to have reduced clearance relative to low-risk type infections in HIV-negative women (HR: 0.57, 95% CI: 0.22–1.45), while HPV18-related types did not (HR: 1.27, 95% CI: 0.74–2.15). Prevalent infections had a longer observed median duration than incident HPV infections (1.4 and 0.6 years, respectively). However, estimated HRs for incident infections were similar to those for prevalent and incident infections combined. When HPV16, HPV16-related, HPV18, HPV18-related, and “other high-risk” types were included in a single model, HRs for clearance relative to low-risk types were similar to the results shown.
|Events1||Woman-years||HR (95% CI)2||Events1||Woman-years||HR (95% CI)2|
|Prevalent and incident infections combined3|
|High-risk6||348||934.3||0.93 (0.79–1.10)||86||138.7||0.95 (0.67–1.35)|
|HPV16 and related7||143||451.3||0.79 (0.64–0.97)||51||91.0||0.86 (0.59–1.27)|
|HPV16||26||77.7||0.85 (0.56–1.30)||16||33.7||0.94 (0.55–1.63)|
|HPV16–related7||117||373.6||0.77 (0.62–0.97)||35||57.3||0.85 (0.55–1.29)|
|HPV18 and related8||85||236.8||0.86 (0.67–1.11)||20||28.0||1.01 (0.62–1.64)|
|HPV18||29||85.3||0.80 (0.56–1.16)||4||9.6||0.57 (0.22–1.45)|
|HPV18–related8||56||151.4||0.89 (0.67–1.20)||16||18.4||1.27 (0.74–2.15)|
|“Other high-risk”10||120||246.2||1.24 (0.99–1.54)||15||19.7||1.18 (0.65–2.15)|
|Incident infections only3|
|High-risk6||236||205.2||0.94 (0.77–1.15)||61||45.2||0.99 (0.66–1.48)|
|HPV16 and related7||92||82.4||0.80 (0.62–1.04)||34||27.9||0.81 (0.51–1.28)|
|HPV16||18||16.7||0.88 (0.55–1.40)||11||8.8||1.06 (0.58–1.93)|
|HPV16-related7||74||65.7||0.78 (0.59–1.03)||23||18.8||0.74 (0.45–1.22)|
|HPV18 and related8||54||45.7||0.87 (0.64–1.20)||14||9.6||1.07 (0.59–1.94)|
|HPV18||17||15.0||0.73 (0.44–1.21)||4||4.2||0.56 (0.21–1.50)|
|HPV18-related8||37||30.8||0.95 (0.67–1.35)||10||5.4||1.67 (0.84–3.29)|
|“Other high- risk”9||90||77.0||1.20 (0.93–1.55)||13||7.9||1.21 (0.62–2.35)|
Among HIV-positive women, estimated clearance rates for infections with HPV16 and related types were also relatively low when compared with infections by all other high-risk types combined (HR: 0.77, 95% CI: 0.62–0.95) and all other HPV types combined (HR: 0.73, 95% CI: 0.61–0.87), while estimated clearance rates for infections with HPV18 and related types were similar to those for all other high-risk types combined (HR: 0.92, 95% CI: 0.72–1.18), but reduced relative to all other HPV types combined (HR: 0.83, 95% CI: 0.66–1.03). Estimated clearance rates for infections with HPV16 and related types among HIV-negative women also were low relative to all other high-risk types combined (HR: 0.75, 95% CI: 0.49–1.15) and all other HPV types combined (HR: 0.84, 95% CI: 0.62–1.13), while estimates for infections with HPV18 and related types were comparable to both reference groups. Clearance for HPV16 and related type infections did not differ greatly from that of HPV18 and related type infections in HIV-positive women (HR: 0.93, 95% CI: 0.71–1.22). The HR was 0.85 (95% CI: 0.52–1.41) in HIV-negative women.
The independent effect of HIV serostatus on HPV clearance was estimated by comparing the clearance of low-risk HPV infections among HIV-positive versus -negative women. The independent effect of high-risk HPV infections was estimated by comparing clearance of high-risk versus low-risk infections among HIV-negative women. The association between HIV-serostatus and estimated clearance (HR 0.63, 95% CI: 0.45–0.88) was stronger than that between high-risk HPV type and clearance (HR: 0.94, 95% CI: 0.67–1.32) (Table III). However, the estimate for the combined effects of HIV positivity and high-risk HPV type was consistent with expectations for multiplicative effects on clearance (observed HR: 0.58, 95% CI: 0.42–0.81 for clearance of high-risk infections among HIV-positive women relative to clearance of low-risk infections among HIV-negative women; expected HR for multiplicative effects = 0.59, likelihood ratio test p-value = 0.92). Although CD4 count was associated with HPV clearance in HIV-positive women (HR for 200–500 cells/μL versus 500+ cells/μL: 0.76, 95% CI: 0.64–0.90; HR for <200 cells/μL versus 500+ cells/μL: 0.47, 95% CI: 0.37–0.58), CD4 count was not associated with high-risk versus low-risk HPV type (exposure odds ratio (eOR) for 200–500 cells/μL versus 500+ cells/μL: 1.16, 95% CI: 0.86–1.56; eOR for <200 cells/μL versus 500+ cells/μL: 1.06, 95% CI: 0.76–1.47). Thus, stratifying by CD4 count did not change the lack of interaction between HIV positivity and high-risk HPV type on the multiplicative scale (likelihood ratio test p-value = 0.37). Results were similar when joint and independent effects on clearance were estimated for HIV serostatus and infection with HPV16 and related types relative to low-risk HPV types and for HIV serostatus and infection with HPV18 and related types relative to low-risk HPV types.
|Sub-population||HR (95% CI)1||LRT p-value2|
|High-risk HPV and HIV interaction|
|LR3 HPV in HIV-negative women||1.004||0.92|
|HR5 HPV in HIV-negative women||0.94 (0.67–1.32)|
|LR3 HPV in HIV-positive women||0.63 (0.45–0.88)|
|HR5 HPV in HIV-positive women||0.58 (0.42–0.81)|
|HPV16 and related types and HIV interaction|
|LR3 HPV in HIV-negative women||1.004||0.67|
|HPV16 and related6 in HIV-negative women||0.86 (0.59–1.26)|
|LR3 HPV in HIV-positive women||0.64 (0.46–0.90)|
|HPV16 and related6 in HIV-positive women||0.50 (0.36–0.71)|
|HPV18 and related types and HIV interaction|
|LR3 HPV in HIV-negative women||1.004||0.84|
|HPV18 and related7 in HIV-negative women||1.02 (0.63–1.64)|
|LR3 HPV in HIV-positive women||0.64 (0.45–0.89)|
|HPV18 and related7 in HIV-positive women||0.54 (0.38–0.79)|
This study of the natural history of type-specific HPV clearance is one of the first studies to examine the clearance of individual HPV types and genetically related types in both HIV-seropositive and HIV-seronegative women. None of the covariates, including ethnicity, age, CD4 count, and other common variables of concern, was an important confounder of the association between HPV type and HPV clearance. Our results suggested a longer duration of HPV16 and related type infections relative to other HPV types and to other high-risk HPV types. However, a difference in estimated clearance rates for HPV16 and related types relative to HPV18 and related types was not seen. In addition, while our results were consistent with previous studies that found a longer duration of HPV infections in HIV-seropositive than HIV-seronegative women,18, 32 there was little evidence that HIV serostatus modified associations between HPV type and HPV clearance.
To our knowledge, only one published study has compared PCR-based clearance rates for specific HPV types or related HPV types compared with clearance of low-risk types. That cohort included 227 cytologically normal Colombian women who were positive for HPV at baseline and of which the majority were HIV-negative.21 Lower clearance rates for HPV16 and HPV16-related types versus low-risk types were noted in both studies. However, the magnitude of these associations among HIV-negative women in the current study (HR: 0.94, 95% CI: 0.55–1.63 for HPV16; HR: 0.85, 95% CI: 0.55–1.29 for HPV16-related types) were weaker than those reported for the Colombian study (HR: 0.47, 95% CI: 0.31–0.69 for HPV16; HR: 0.67, 95% CI: 0.47–0.94 for HPV16-related types). Estimated HRs for HPV18 and related types suggested that clearance rates were comparable to low-risk types in the Colombian study population (HR: 0.98, 95% CI: 0.70–1.39) and among HIV-negative women in HERS (HR: 1.01, 95% CI: 0.62–1.64), although clearance for HPV18 alone was reduced relative to low-risk type infections that occurred in HIV-negative women in HERS (HR: 0.57, 95% CI: 0.22–1.45). Clearance rates for “other high-risk” HPV types were slightly increased relative to low-risk types for HIV-negative women in the current study (HR: 1.18, 95% CI 0.65–2.15) and slightly decreased for women in the Colombian study (HR: 0.87, 95% CI: 0.55–1.36). Both studies defined the “other high-risk” category as all high-risk HPV types not phylogenetically related to HPV16 or 18, but the specific HPV types detected within this other high risk category differed (HPV26, 51, 56, 66, 73, and 82 in the current study; HPV51, 56, and 66 in the Colombian study).
A previous study of 2,929 HIV-seropositive women, including 871 participants in the HERS cohort, found that immune suppression (as measured by reduced CD4 count) was positively associated with a higher incidence and prevalence of different specific HPV types.33 However, the association with HPV16 appeared weak compared with estimates for other HPV types, suggesting that HPV16 may be better able to evade immune responses than other HPV types.33 Similarly, in the current study HIV serostatus did appear to influence the clearance of HPV infection; the median duration of HPV infection was longer in HIV-positive than HIV-negative women for most HPV types, but not for HPV16. In the time to clearance analysis, however, HIV infection did not appear to substantially modify the effect of high-risk HPV type, HPV16 and related types, or HPV18 and related types on HPV clearance, even when stratified by CD4 count.
The PCR-based evaluation of type-specific HPV infections is a strength of this study, as each HPV type is genetically distinct and was considered as a separate infection.34 This study also included a larger number of type-specific HPV infections than the only previous study of specific HPV type clearance,21 which allowed an independent evaluation of HPV18. Another strength was our ability to stratify by HIV serostatus, although the number of type-specific HPV infections was somewhat limited in HIV-seronegative women.
As in all longitudinal studies of HPV infection, the reported mean and median durations of HPV infections are likely to be underestimates due to left-censoring of infections present at the start of the study and right-censoring of infections that did not clear by the end of follow-up. This underestimation of the duration of HPV infections might attenuate differences in associations of high-risk types with HPV clearance if differences in duration are only apparent after an extended period of infection (for example, after at least a year of infection). However, HRs by HPV type or category for prevalent and incident HPV infections combined were similar to those for incident HPV infections alone.
The enrollment criteria for the HERS cohort (HIV seropositivity or high-risk sexual behavior or injection drug use) may limit generalizability to other populations. The smaller number of HPV infections in HIV-negative women also led to imprecision in some estimates. However, the median duration of HPV infections among HIV-negative women in HERS, the most appropriate comparison to other studies in largely HIV-negative populations, were generally consistent with estimates based on other study populations (0.6 years compared with 0.5,21 0.7,3 and 1.4 years20 for all HPV; 0.6 years compared with 0.711 and 1.1 years20 for high-risk HPV; 0.5 years compared with 0.411 and 1.0 years20 for low-risk HPV; and 1.0 year compared with 0.7,11 0.9,3, 19 and 1.6 years20 for HPV16). Thus, the median duration of HPV infections occurring in HIV-negative women in HERS is comparable to those reported in other populations.
While HPV detection was conducted in a central laboratory with sensitive PCR primers capable of amplifying a wide range of HPV types, a large proportion of HPV infections could not be typed. Currently, there are primer sets available that are more sensitive for detecting multiple HPV types.35 For example, in a study of 262 cervicovaginal lavage specimens, the PGMY-line blot assay detected multiple HPV types in 81 (46.0%) of 176 HPV-positive samples, while the MY09/MY11 primer system detected multiple HPV types in only 69 (43.7%) of 158 HPV-positive samples.35 It is therefore possible that some HPV infections in samples with multiple infections were missed. If the decreased sensitivity of the MY09/MY11 primers led to single false negative visits, however, the requirement for clearance of an HPV infection to have been confirmed by 2 consecutive negative visits is likely to have reduced this potential misclassification.
In summary, HPV16 and related infections had low estimated clearance rates relative to low-risk HPV infections. There was also some suggestion that HPV18 may be associated with lower HPV clearance relative to low-risk types. Despite the clear association between HIV serostatus and HPV clearance, no substantial variation was noted in estimated HRs for HPV type and clearance by HIV serostatus. Although the large study sample enabled independent examinations of the clearance of HPV16 and HPV18 infections, it would also be desirable to separately examine additional individual HPV types, notably those in the “other high-risk” category (HPV26, 51, 56, 66, 73, 82). Therefore, future studies should ideally include a larger number of women and evaluate a wider range of specific HPV types to further clarify the association between HPV persistence and specific HPV types.
The authors thank Ms. Pangaja Paramsothy for assisting with HERS data and Dr. Andrew Olshan for assisting in project design and manuscript review. The authors also thank the additional HERS Study Group Members: Julia Arnsten, MD, MPH, Robert D. Burk, MD, Chee Jen Chang, PhD, Penelope Demas, PhD, and Andrea Howard, MD, MSc, Ellie Schoenbaum, MD, Montefiore Medical Center and the Albert Einstein College of Medicine, Bronx, New York; Jack Sobel, MD, Wayne State University School of Medicine, Detroit, Michigan; Anne Rompalo, MD, and David Vlahov, PhD, Johns Hopkins University School of Hygiene and Public Health, Baltimore, Maryland; Charles Carpenter, MD, Timothy Flanigan, MD, Joseph Hogan, Kenneth Mayer, MD, ScD, Josiah Rich, MD, Valerie Stone, MD, and Karen Tashima, MD, Brown University School of Medicine, Providence, Rhode Island; Lytt I. Gardner, PhD, Scott D. Homberg, MD, Janet S. Moore, PhD, Ruby M. Phelps, BS, Dawn K. Smith, MD, MPH, and Dora Warren, PhD, Centers for Disease Control and Prevention, Atlanta, Georgia; and Katherine Davenny, MPH, National Institute of Drug Abuse, Rockville, Maryland. Jill Koshiol is now a Cancer Prevention Fellow in the Cancer Prevention Fellowship Program, Division of Cancer Prevention, National Cancer Institute, National Institutes of Health, DHHS, Bethesda, MD 20892.
- 1GLOBOCAN 2002: cancer incidence, mortality and prevalence worldwide, Version 2.0, vol. 2005. IARC CancerBase No. 5. Lyon: IARC Press, 2004., , , .
- 25Centers for Disease Control. Revision of the CDC surveillance case definition for acquired immunodeficiency syndrome. Morbid Mortal Wkly Rep. 1987; 36: 1S–15S.
- 28Approximate statistics: the likelihood-ratio method. In: RothmanKJ, GreenlandS, eds. Modern epidemiology, 2nd edn. Philadelphia, PA: Lippincott Williams and Wilkins, 1998. 218–20., .
- 30Repeated events. In: Survival analysis using the SAS system: a practical guide. Cary, NC: SAS Institute Inc, 1995. 236–47..
- 31Cox models with nonproportional hazards. In: Survival analysis using the SAS system: a practical guide. Cary, NC: SAS Institute, 1995. 154–7..