Abstract

Background: Women infected with human immunodeficiency virus type 1 (HIV-1) and -2 may be at higher risk of developing cervical cancer than uninfected women. We assessed the relationships among human papillomavirus (HPV) types and persistence, HIV-1 and/or HIV-2 infection, and the development of high-grade cervical squamous intraepithelial lesions (HSILs) in a prospective study. Methods: We studied 627 women with and without HIV-1 and/or HIV-2 infection and high-risk HPV infection in Senegal, West Africa, who were assessed every 4 months for HSIL and HPV DNA over a mean follow-up of 2.2 years. Cox regression modeling was used to assess risks associated with development of HSIL. Results: During follow-up, 71 (11%) of 627 women developed HSIL as detected by cytology. HIV-infected women with high-risk HPV types were at greatest risk for development of HSIL. In multivariable modeling, infection with oncogenic HPV types—both persistent (hazard ratio [HR] = 47.1, 95% confidence interval [CI] = 16.3 to 136) and transient (HR = 14.0, 95% CI = 3.7 to 54)—was strongly associated with HSIL risk. In univariate analyses, HIV-positive women infected with HIV-2 were less likely to develop HSIL (HR = 0.3, 95% CI = 0.1 to 0.9) than HIV-positive women infected with HIV-1. HIV-positive women with CD4 + cell counts between 200 and 500 cells per microliter (HR = 2.2, 95% CI = 0.8 to 6.3) or fewer than 200 cells per milliliter (HR = 5.5, 95% CI = 2.0 to 15.2) were at greater risk of HSIL than HIV-positive women with CD4 counts of more than 500 cells per milliliter. High plasma HIV RNA levels were associated with increased HSIL risk (HR for each order of magnitude increase in the level of plasma HIV RNA = 1.4, 95% CI = 1.1 to 1.7; P = .005). After adjustment for HPV types and persistence, however, HIV type, plasma HIV RNA level, and CD4 count were no longer statistically significantly associated with increased risk of HSIL. Conclusions: HIV-1 and HIV-2 are associated with increased risk for development of HSIL. This risk appears to be associated primarily with increased HPV persistence that may result from immunosuppression related to HIV-1 and/or HIV-2 infection.

Infection with high-risk types of human papillomavirus (HPV) is required for the development of invasive cervical cancer ( 14 ) . HPV oncoproteins E6 and E7 set the stage for development of malignancy by disrupting normal host cell cycle controls by binding to and promoting the degradation of p53 by increasing the expression of telomerase (E6) and by binding to the retinoblastoma protein (E7). Our studies ( 57 ) and those of others ( 811 ) have shown that human immunodeficiency virus (HIV) infection is associated with an increased risk for detection of high-risk types of HPV, cervical cancer, and the cervical cancer precursor lesions low-grade and high-grade squamous intraepithelial lesions (LSILs and HSILs, respectively). In a previous cross-sectional study among women infected with high-risk types of HPV ( 7 ) , women with HIV-1, HIV-2, and dual HIV-1 and HIV-2 infection were 2.2, 6.0, and 8.0 times, respectively, more likely to have HSIL and invasive cancer than those without HIV infection.

The mechanism by which HIV infection increases risk of cervical neoplasia is not completely understood. One plausible explanation is that HIV-induced immunosuppression leads to an inability to control the expression of HPV and the production of HPV oncoproteins E6 and E7. Several previous longitudinal studies have assessed the effect of HIV-1 infection, HIV viral load, and HIV-associated immunosuppression on the detection of high-risk types of HPV and on the risk of cervical neoplasia ( 1216 ) . However, in most of those studies women were monitored only to the development of LSIL, a lesion that is generally transient and associated with a low risk of neoplastic progression, rather than to development of HSIL or carcinoma in situ, lesions with substantial neoplastic potential. Moreover, no data are available that describe the risk of developing HSIL or carcinoma in situ associated with HIV-2 infection. We therefore undertook this study to characterize the risk of developing HSIL/carcinoma in situ that is associated with HPV types and persistence, HIV-1 or HIV-2 infection, CD4 + cell counts, and HIV viral load.

M ATERIALS AND M ETHODS

Study Population

We conducted a prospective study of HSIL development among women with and without HIV-1 and/or HIV-2 infection and high-risk HPV infection in Senegal, West Africa. Between October 1, 1994, and January 1, 1998, all women older than 15 years presenting to the University of Dakar Infectious Disease Clinic (n = 4349) and commercial sex workers attending one of two sexually transmitted disease clinics in Dakar (n = 773) and in M'Bour (n = 270) were offered serologic testing for HIV-1 and HIV-2 and screening for cervical HPV DNA and cytology. Women returned 4 weeks later for their HIV, HPV, and cervical cytology results, and those found to be HIV seropositive were counseled according to the guidelines established by the Senegalese AIDS national committee. None of the HIV-infected women was receiving antiretroviral therapy. We invited all women with HIV or high-risk HPV infection to participate in a longitudinal study, with visits occurring every 4 months, to assess the risk of developing cervical lesions. Of the 5392 women screened, 1348 (25%) were positive for HIV-1 and/or HIV-2 or had high-risk HPV types detected in cervical swabs. Of these 1348 eligible women, 939 (70%) enrolled. Enrollment among HIV-positive and/or HPV-positive women varied by site of recruitment, with 63%, 85%, and 88% of such women from the University of Dakar Outpatient Infectious Disease Clinic, the Dakar sexually transmitted disease clinic, and the M'Bour sexually transmitted disease clinic, respectively, consenting to enroll. Enrollment rates varied somewhat by HIV status; 76% of HIV-positive women (with or without HPV infection) and 64% of HPV-positive, HIV-negative women enrolled. However, after adjusting for study site and HIV status, enrollment rates did not differ by demographic or behavioral characteristics, such as age, marital status, parity, use of contraception, or the use of alcohol or tobacco. In addition to HIV-positive and/or HPV-positive women, a subset of 177 HIV-negative women without HPV infection (i.e., 4% of the 3937 without HIV or HPV infection) were also enrolled in the study as a reference group at low risk of developing HSIL during follow-up. Written informed consent was obtained from all enrolled women. These women were screened earlier in the study, were more likely to be from one of the two sexually transmitted disease clinics, and were somewhat older than the HIV-negative, HPV-negative women who were not enrolled. They, however, did not differ by other measured demographic or behavioral factors after adjustment for study site and age. In total, the longitudinal study enrolled 1116 women. The study was conducted according to procedures approved by the institutional review boards of both the University of Washington and the University of Dakar.

Collection of Specimens and Study Procedures

At the screening visit, blood was collected for HIV-1 and HIV-2 serologic assays, and cervical cellular samples were obtained for cytologic and HPV screening as previously described ( 7 ) . Subjects underwent a general physical examination and completed a short standardized interview, including questions pertaining to medical history and to sexual behavior. At the return visit 4 weeks later, and at each subsequent 4-month follow-up visit, enrolled subjects underwent a more detailed interview regarding sexual behavior that included questions about marital history, contraception use, age at first intercourse, and number of lifetime sexual partners. Interviewers also asked questions about medical history to obtain information about previous hospitalizations and the use of medications. At each visit, blood was obtained for qualitative and quantitative RNA and DNA assays of HIV-1 and/or HIV-2 and for lymphocyte subset analysis.

Cytology Screening

Pap smears were interpreted and classified according to the Bethesda System ( 17 ) as unsatisfactory, negative, atypical squamous cells of uncertain significance, LSIL, HSIL, or invasive cervical cancer. From October 1, 1994, through March 31, 1998, conventional Pap smears were used. Slides were initially read by a pathologist in Senegal and then sent to Seattle for a second reading by a cytotechnologist and the study cytopathologist (N.B.K.), who were blinded to HIV and HPV results. All conventional slides classified as LSIL or worse by either pathologist were restained, recovered, and reread by the pathologist in Seattle. The final cytologic diagnosis used in our analyses was that made by the Seattle pathologist; however, if the slide was unavailable (broken or lost) for reexamination in Seattle, the diagnosis made by the Senegalese pathologist was used. Beginning April 1, 1998, the Thin Prep monolayer cell preparation system (Cytyc Corp., Boxborough, MA) was used to obtain all Pap smears, and diagnoses from these slides were made by the cytopathologist in Seattle.

HPV DNA Detection and Typing

Polymerase chain reaction (PCR) assays for the detection of HPV DNA were performed with HPV L1 consensus primers and HPV type-specific oligonucleotide probes specific for high-risk HPV types 16, 18, 31, 33, 35, 39, 45, 51, 52, or 56 and a generic HPV probe, as previously described ( 18 ) . Initially, screening for the presence of high-risk HPV was done with a 10-probe mixture, testing for HPV DNA by PCR that used the consensus primers MY09 and MY11, which are specific for a highly conserved region in the L1 open reading frame. Positive samples were then reamplified to assess presence of 12 HPV types in primer groups for low-risk HPV types (i.e., combined HPV 6 and 11) and for high-risk HPV types (i.e., HPV 16; HPV 18; combined HPV 31, 33, 35, and 39; combined HPV 45 and 56; and combined HPV 51 and 52). Beginning April 1, 1998, when new probes were available, HPV detection and typing analyses were performed via a PCR-based reverse-line strip test method (Roche Molecular Systems, Alameda, CA) with probes for low-risk HPV types 6, 11, 40, 42, 53, 54, 57, 66, and 84 and high-risk HPV types 16, 18, 26, 31, 33, 35, 39, 45, 51, 52, 55, 56, 58, 59, 68, 73, 82, and 83, as previously described ( 19 ) .

HIV Serology and Lymphocyte Subset Analysis

Initial screening of serum samples for HIV-1 or HIV-2 antibodies was performed with a microwell plate enzyme immunoassay that detects antibodies to both HIV types in one well (HIV-1/2 EIA; Sanofi Diagnostics Pasteur, Redmond, WA). Positive samples were confirmed with a rapid, HIV peptide–based membrane immunoassay that distinguishes between antibodies to HIV-1 and HIV-2 (Multispot; Genetic Systems, Redmond, WA). Whole blood collected in EDTA tubes was analyzed by flow cytometry with a FACSCount analyzer (Becton Dickinson Biosciences, San Jose, CA) to determine the number of CD4 + , CD8 + , and CD3 + cells per microliter of blood. Cell counts were performed for all samples from HIV-infected women and for all baseline visit samples from HIV-negative women.

Quantitation of HIV-1 and HIV-2 Plasma RNA

Quantitative and qualitative assays for HIV-1 RNA were performed as previously described ( 20 ) . The qualitative assay was performed on all available samples that were negative by the quantitative assay. The quantitative and qualitative HIV-1 RNA assays detected as few as 80 and 40 copies of HIV-1 RNA per milliliter, respectively, with reproducible sensitivities of 400 and 200 HIV-1 RNA copies per milliliter, respectively. The HIV-2 quantitative and qualitative assays detect as few as 40 and 20 copies of HIV-2 RNA per milliliter, respectively, with reproducible sensitivities of 200 and 100 HIV-2 RNA copies per milliliter, respectively.

Statistical Methods

Cervical HPV DNA types were categorized according to their known associations with invasive cervical cancer, with high-risk types including HPV 16, 18, 31, 33, 35, 39, 45, 51, 52, and 56 and with low-risk types including HPV 6 and 11. Untyped/other HPV types were those HPV types that were included in the reverse-line strip test but not the 12-type mixture used before 1998 (HPV types 26, 40, 42, 53, 54, 55, 56, 57, 58, 59, 66, 68, 73, 82, 83, and 84) or unclassified HPV types. HPV oncogenicity grouping was hierarchical, and categories were mutually exclusive. A sample was called “HPV negative” if HPV was not detected by any of the methods used. At a given study visit, persistence of a “high-risk,” “low-risk,” or “untyped/other” HPV type was defined as the detection at the current and immediately preceding study visit (that is, two consecutive visits) of an HPV type within the indicated risk group, regardless of the specific type detected. At a given study visit, a current HPV infection was considered transient if it was not detected at the immediately preceding visit (that is, if HPV was not detected at two consecutive visits).

Women with initial cytologic diagnoses of LSIL or worse were excluded from analysis. At least two visits with a satisfactory Pap smear (i.e., with a sufficient squamous component and less than 75% of epithelial cells obscured) were necessary for inclusion into the analysis. For women who had an unsatisfactory Pap smear at the enrollment visit, the first satisfactory negative or atypical Pap smear after the initial unsatisfactory Pap smear(s) was considered as the baseline visit. Incidence rates for development of HSIL, as determined by cytology, stratified by HIV status and high-risk HPV detection at the baseline visit, were estimated by Kaplan–Meier methods ( 21 ).

Cox proportional hazards regression analyses, with time-dependent covariates, were performed to evaluate the independent effects of HIV status and HPV type (independent variables) on time to development of HSIL (outcome), adjusting for the number of follow-up visits and other confounding factors. The proportional hazards assumption was tested by creating and assessing the significance of interaction terms for all predictors of interest and our time variable. All final models met the proportional hazards assumption. To investigate any potential lack of fit of our model, we conducted a formal analysis of residuals. Martingale (for continuous factors such as HIV plasma load) and deviance residuals for the proportional hazards regression analyses were plotted to look for outliers. No overly influential outliers were identified.

In analyses of the entire study population, women without detectable HPV were used as the reference group. However, in subgroup analyses among HIV-positive women, because of the lack of HSIL development in this reference group, women with transient or persistent HPV 6 or HPV 11 infection and transient untyped/other HPV infection, in addition to women with no HPV infection, were used as a reference category. Relative risks (RRs) were estimated by use of hazard ratios. Factors examined as potential confounders included employment as a commercial sex worker, age, marital status, parity, contraceptive method, smoking, and alcohol use; factors were retained in a specific model if the unadjusted and adjusted coefficients for HIV status (HPV type) differed by more than 10% ( 22 ) . A priori, CD4 counts were stratified into three groups (>500 cells per microliter, 200–500 cells per microliter, and <200 cells per microliter) and plasma HIV RNA levels were evaluated per order of magnitude increase in HIV-1 or HIV-2, as in previous analyses ( 7 , 20 ) .

Factors measured at baseline and follow-up visits (e.g., presence and persistence of types of HPV DNA, new sex partners, CD4 count, and HIV plasma RNA level) were included as time-dependent variables. In this manner, for a given subject, incident HPV infections, HPV persistence, and clearance of various HPV types were accounted for and evaluated over time. This procedure was important because the median length of follow-up in this study was longer than the estimated average time of detection of prevalent (i.e., 6–12 months) ( 23 , 24 ) or incident (i.e., 1–2 years) ( 25 ) HPV infections described in previous studies. Sensitivity analyses were conducted to determine whether the differences in the definition of persistence substantially altered study findings.

Among HIV-positive subjects, Cox regression analyses separately evaluating HIV type, CD4 counts, plasma HIV RNA levels, and HPV detection and persistence were performed because these factors were associated with each other and have been previously shown to be associated with cervical neoplasia ( 7 ) . To determine the potential independent association of each factor with development of HSIL after simultaneously controlling for the effects of the other factors, a final composite Cox regression analysis was performed. Data analyses were performed with SAS version 9.1 (SAS Institute, Cary, NC). All statistical tests were two-sided.

R ESULTS

Characteristics of Study Population

Of 1116 women enrolled (i.e., 939 positive for HIV and/or HPV and 177 negative for both HIV and HPV), 137 with an initial cytologic diagnosis of LSIL or worse and 17 who never had a satisfactory Pap smear were not included in these analyses. Of the remaining 962 women, 627 (246 HIV-positive women and 381 HIV-negative women) had at least one follow-up visit for which a satisfactory Pap smear was available and were thus included in the final analyses. HIV-positive women and HIV-negative women included in our analyses were similar with regard to site of recruitment, parity status, smoking behavior, alcohol use, age at first sexual intercourse, employment as a commercial sex worker, and length of follow-up ( Table 1 ). However, HIV-positive women, compared with HIV-negative women, were more likely to be older; to be divorced, separated, or widowed; to have been born outside of Senegal; to not use contraception; and to have had more follow-up visits. Among HIV-positive women, those infected with HIV-2 were older and more likely to be a commercial sex worker than those infected with HIV-1.

Table 1.

Demographic and laboratory characteristics of the study population by human immunodeficiency virus (HIV) status at the enrollment visit *

  HIV status
 
     
Characteristic Negative (n = 381)  HIV-1 + (n = 165)   HIV-2 + (n = 57)  Dually infected (n = 24) P P 
Site of recruitment, No. (%)     .3 .002 
    Dakar STD Clinic 108 (28) 36 (22) 15 (26) 12 (50)   
    M'Bour STD Clinic 39 (10) 18 (11) 14 (25) 3 (13)   
    Dakar Outpatient ID Clinic 234 (62) 111 (67) 28 (49) 9 (37)   
Mean age, y (±SD) 30.5 ± 7.4 31.0 ± 7.7 34.2 ± 7.1 32.6 ± 6.3 .02 .02 
Marital status, No. (%)     <.001 .3 
    Never married 81 (21) 23 (14) 7 (12) 4 (17)   
    Married—monogamous 128 (34) 47 (28) 17 (30) 3 (13)   
    Married—polygamous 52 (14) 19 (11) 4 (7) 5 (21)   
    Divorced/separated 113 (29) 48 (29) 24 (42) 10 (42)   
    Widowed 6 (2) 28 (17) 5 (9) 2 (8)   
Country of birth—Senegal, No. (%) 339 (89) 130 (79) 51 (89) 20 (83) .006 .2 
Parity ≥ 1, No. (%) 333 (88) 139 (85) 52 (91) 21 (88) .7 .5 
Lack of contraception use, No. (%) 139 (37) 109 (66) 24 (43) 9 (39) <.001 .02 
Current cigarette use, No. (%) 91 (24) 34 (21) 20 (35) 7 (30) .8 .06 
Current alcohol use, No. (%) 49 (13) 25 (15) 13 (23) 7 (29) .07 .13 
Mean age at first intercourse, y (±SD) 17.5 ± 3.6 17.2 ± 3.3 16.4 ± 2.9 17.6 ± 4.1 .09 .3 
Commercial sex worker, No. (%) 151 (40) 62 (38) 30 (53) 15 (63) .4 .02 
Cervical HPV DNA detected, No. (%)     <.001 .3 
    No HPV 45 (12) 75 (45) 25 (45) 12 (52)   
    Untyped HPV or HPV 6/11 23 (6) 23 (14) 14 (25) 3 (13)   
    High-risk HPV 310 (81) 67 (41) 17 (30) 8 (33)   
Mean CD4 count, cells per mL (±SD) 910 ± 346 402 ± 249 639 ± 331 583 ± 322 <.001 <.001 
CD4 count, No. (%)       
    >500 cells per mL 321 (91) 53 (32) 37 (65) 12 (52)   
    200–500 cells per mL 29 (8) 72 (43) 15 (26) 9 (39)   
    <200 cells per mL 4 (1) 42 (25) 5 (9) 2 (9)   
Mean CD8 + cell count, cells   per mL (±SD)  488 ± 236 952 ± 501 694 ± 356 909 ± 451 <.001 .002 
Mean log 10 plasma HIV-1 RNA,   copies per mL (±SD)  — 4.24 ± 1.61 — 2.92 ± 2.23 — <.001 
Mean log 10 plasma HIV-2 RNA,   copies per mL (±SD)  — — 1.60 ± 1.60 1.47 ± 1.52 — .4 
Mean duration of follow-up, y (±SD) 2.2 v 1.2 2.2 ± 1.4 2.7 ± 1.6 2.1 ± 1.3 .5 .03 
Mean No. follow-up visits (±SD) 4.6 ± 3.1 5.8 ± 4.3 6.4 ± 4.4 4.7 ± 3.3 <.001 .2 
  HIV status
 
     
Characteristic Negative (n = 381)  HIV-1 + (n = 165)   HIV-2 + (n = 57)  Dually infected (n = 24) P P 
Site of recruitment, No. (%)     .3 .002 
    Dakar STD Clinic 108 (28) 36 (22) 15 (26) 12 (50)   
    M'Bour STD Clinic 39 (10) 18 (11) 14 (25) 3 (13)   
    Dakar Outpatient ID Clinic 234 (62) 111 (67) 28 (49) 9 (37)   
Mean age, y (±SD) 30.5 ± 7.4 31.0 ± 7.7 34.2 ± 7.1 32.6 ± 6.3 .02 .02 
Marital status, No. (%)     <.001 .3 
    Never married 81 (21) 23 (14) 7 (12) 4 (17)   
    Married—monogamous 128 (34) 47 (28) 17 (30) 3 (13)   
    Married—polygamous 52 (14) 19 (11) 4 (7) 5 (21)   
    Divorced/separated 113 (29) 48 (29) 24 (42) 10 (42)   
    Widowed 6 (2) 28 (17) 5 (9) 2 (8)   
Country of birth—Senegal, No. (%) 339 (89) 130 (79) 51 (89) 20 (83) .006 .2 
Parity ≥ 1, No. (%) 333 (88) 139 (85) 52 (91) 21 (88) .7 .5 
Lack of contraception use, No. (%) 139 (37) 109 (66) 24 (43) 9 (39) <.001 .02 
Current cigarette use, No. (%) 91 (24) 34 (21) 20 (35) 7 (30) .8 .06 
Current alcohol use, No. (%) 49 (13) 25 (15) 13 (23) 7 (29) .07 .13 
Mean age at first intercourse, y (±SD) 17.5 ± 3.6 17.2 ± 3.3 16.4 ± 2.9 17.6 ± 4.1 .09 .3 
Commercial sex worker, No. (%) 151 (40) 62 (38) 30 (53) 15 (63) .4 .02 
Cervical HPV DNA detected, No. (%)     <.001 .3 
    No HPV 45 (12) 75 (45) 25 (45) 12 (52)   
    Untyped HPV or HPV 6/11 23 (6) 23 (14) 14 (25) 3 (13)   
    High-risk HPV 310 (81) 67 (41) 17 (30) 8 (33)   
Mean CD4 count, cells per mL (±SD) 910 ± 346 402 ± 249 639 ± 331 583 ± 322 <.001 <.001 
CD4 count, No. (%)       
    >500 cells per mL 321 (91) 53 (32) 37 (65) 12 (52)   
    200–500 cells per mL 29 (8) 72 (43) 15 (26) 9 (39)   
    <200 cells per mL 4 (1) 42 (25) 5 (9) 2 (9)   
Mean CD8 + cell count, cells   per mL (±SD)  488 ± 236 952 ± 501 694 ± 356 909 ± 451 <.001 .002 
Mean log 10 plasma HIV-1 RNA,   copies per mL (±SD)  — 4.24 ± 1.61 — 2.92 ± 2.23 — <.001 
Mean log 10 plasma HIV-2 RNA,   copies per mL (±SD)  — — 1.60 ± 1.60 1.47 ± 1.52 — .4 
Mean duration of follow-up, y (±SD) 2.2 v 1.2 2.2 ± 1.4 2.7 ± 1.6 2.1 ± 1.3 .5 .03 
Mean No. follow-up visits (±SD) 4.6 ± 3.1 5.8 ± 4.3 6.4 ± 4.4 4.7 ± 3.3 <.001 .2 
*

STD = sexually transmitted disease; SD = standard deviation; HPV = human papillomavirus; ID = infectious disease.

P value was from a chi-squared test for difference between HIV-negative women and HIV-positive women. All statistical tests were two-sided.

P value was from chi-squared test for any difference among women infected with HIV-1, with HIV-2, or dually with HIV-1 and HIV-2.

By study design, we oversampled HIV-negative women with high-risk HPV infection to enroll a group of women at high risk of developing HSIL. Consequently, at baseline, among 627 study participants, 402 (64%) were infected with high-risk HPV, including 92 (37%) of 246 HIV-infected and 310 (81%) of 381 HIV-uninfected women ( P <.001, chi-squared test). Women infected with HIV-1 and women infected with HIV-2 were similar to each other with regard to HPV infection (i.e., presence of any HPV or a high-risk type). Women infected with HIV-1 had lower CD4 cell counts but higher CD8 cell counts, as well as higher plasma HIV RNA levels, than women infected with HIV-2. Among women infected with HIV-1, plasma HIV-1 RNA levels were lower among women dually infected with HIV-1 and HIV-2 than among women infected with HIV-1 alone ( P <.001); however, among women infected with HIV-2, plasma HIV-2 RNA levels were similar between women dually infected with HIV-2 and HPV and women infected with HIV-2 alone. None of the HIV-infected women in this cohort was receiving antiretroviral therapy.

Study participants had a mean number of 4.7 follow-up visits (maximum of 18 follow-up visits) over 2.2 years (maximum follow-up of 5.8 years), and 84% of them had at least two follow-up visits. Of 627 women included the final analysis, 70 (11.3%) developed HSIL, as detected by cytology during follow-up, and one woman developed invasive cervical cancer (a woman infected with HIV-1 with four previous negative cytologic test results). For analysis purposes, this subject was considered to have developed HSIL.

Risk of Developing HSIL in Relation to HIV and Baseline HPV Status

During follow-up, 71 (11%) of the 627 women developed HSIL as detected by cytology. Kaplan–Meier survival curves were calculated for development of HSIL during follow-up for women without cytologic abnormalities detected at baseline. HIV-positive women coinfected with high-risk HPV types were at greatest risk for development of HSIL ( Fig. 1 ; P <.001). The 71 HIV-negative women without high-risk HPV infection did not develop HSIL within the first 2 years, but by 3 years, 6% of high-risk women who were HPV negative, regardless of HIV status, had developed HSIL. Among the 402 women who were positive for high-risk HPV DNA at study entry, 10.3% (95% confidence interval [CI] = 6% to 14%) of 310 HIV-negative women and 34.1% (95% CI = 20% to 48%) of 92 HIV-positive women developed HSIL by 3 years of follow-up.

Fig. 1.

Cumulative incidence of high-grade cervical squamous intraepithelial lesion (HSIL), by human immunodeficiency virus (HIV) status and detection of high-risk human papillomavirus (HPV) types at baseline. Number of subjects at risk and survival estimates, with 95% confidence intervals in parentheses, are shown at 12, 24, and 36 months of follow-up.

Fig. 1.

Cumulative incidence of high-grade cervical squamous intraepithelial lesion (HSIL), by human immunodeficiency virus (HIV) status and detection of high-risk human papillomavirus (HPV) types at baseline. Number of subjects at risk and survival estimates, with 95% confidence intervals in parentheses, are shown at 12, 24, and 36 months of follow-up.

Risk of incident HSIL among HIV-infected women varied by HIV type ( P = .04), with women infected with only HIV-1 and women infected with both HIV-1 and HIV-2 more likely to develop HSIL during follow-up than women infected with HIV-2 alone ( Fig. 2 ). Among HIV-infected individuals, increased levels of immunosuppression were strongly associated with increased development of HSIL ( Fig. 3 ; P <.001). By 3 years, only 5.7% (95% CI = 0% to 11%) of 102 women with CD4 counts of more than 500 cells per microliter developed HSIL during follow-up, compared with 19.7% (95% CI = 10% to 30%) of 95 women with CD4 counts between 200 and 500 cells per microliter and 38.9% (95% CI = 14% to 64%) of 48 women with CD4 counts of less than 200 cells per microliter.

Fig. 2.

Cumulative incidence of high-grade cervical squamous intraepithelial lesion (HSIL) among human immunodeficiency virus (HIV)–positive women, by HIV type. Number of subjects at risk and survival estimates, with 95% confidence intervals in parentheses, are shown at 12, 24, and 36 months of follow-up.

Fig. 2.

Cumulative incidence of high-grade cervical squamous intraepithelial lesion (HSIL) among human immunodeficiency virus (HIV)–positive women, by HIV type. Number of subjects at risk and survival estimates, with 95% confidence intervals in parentheses, are shown at 12, 24, and 36 months of follow-up.

Fig. 3.

Cumulative incidence of high-grade cervical squamous intraepithelial lesion (HSIL) among human immunodeficiency virus (HIV)–positive women, by baseline CD4 + cell count, expressed as cells per microliter. Number of subjects at risk and survival estimates, with 95% confidence intervals in parentheses, are shown at 12, 24, and 36 months of follow-up.

Fig. 3.

Cumulative incidence of high-grade cervical squamous intraepithelial lesion (HSIL) among human immunodeficiency virus (HIV)–positive women, by baseline CD4 + cell count, expressed as cells per microliter. Number of subjects at risk and survival estimates, with 95% confidence intervals in parentheses, are shown at 12, 24, and 36 months of follow-up.

Multivariable Cox regression analyses were performed to examine the risk of developing HSIL in relation to detection of specific types of HPV and HIV status at study entry after adjusting for age, employment as a commercial sex worker, parity, and birthplace in Senegal. Among women in whom high-risk types of HPV were not detected at study entry, HIV-positive women were not statistically significantly more likely to develop HSIL (hazard ratio [HR] = 3.0, 95% CI = 0.7 to 13.3) than HIV-negative women (data not shown). However, among HIV-negative women, those with high-risk types of HPV detected at baseline were 4.3 times more likely to develop HSIL during follow up (HR = 4.3, 95% CI = 1.0 to 18.1) than those who were negative for high-risk HPV at study entry; HIV-positive women with high-risk HPV detected at baseline were more than 11 times more likely to develop HSIL during follow-up (HR = 11.5, 95% CI = 2.7 to 49) than HIV-negative women without high-risk HPV at baseline. Among women with high-risk HPV detected at baseline, those infected with HIV-1 were at elevated risk for developing HSIL (HR = 3.0, 95% CI = 1.6 to 5.6) than women without HIV infection, as were women infected with HIV-1 and HIV-2 (HR = 5.9, 95% CI = 1.7 to 20.8), whereas women infected with HIV-2 alone were not at increased risk (HR = 1.2, 95% CI = 0.3 to 3.9).

Association of HIV infection, Transient or Persistent Detection of HPV DNA, and Development of HSIL

Of the 402 women who were infected with high-risk HPV (types 16, 18, 31, 33, 35, 39, or 45) at baseline, 164 (41%) had persistent high-risk HPV, defined as detection of any high-risk type of HPV DNA in at least two consecutive follow-up visits. HIV-infected women were statistically significantly more likely than HIV-negative women to have persistent detection of high-risk types of HPV (61% versus 35%, respectively; P <.001). Among HIV-infected women with high-risk HPV at baseline, persistent detection of high-risk HPV did not vary substantially by HIV type, because women infected with HIV-1 and with high-risk HPV at baseline were somewhat more likely than women infected with HIV-2, but less likely than women infected with both HIV-1 and HIV-2, to have persistent detection of HPV during follow-up (63% versus 44% versus 88%, respectively; P = .10). Persistent detection of untyped or other HPV types was also more common in HIV-infected women than in HIV-negative women during follow-up (38% versus 23%, respectively; P = .004).

Multivariable Cox regression analyses with time-dependent covariates were performed to examine the risk of developing HSIL in relation to HPV type and persistence (both of which were allowed to vary at each time point during follow-up) and to HIV status. In all models, risk estimates were adjusted for baseline age, parity, birthplace in Senegal or elsewhere, and commercial sex work, as well as cytology reader and type of Pap smear (conventional versus Thin Prep), which were also allowed to vary over time. Compared with women without an HPV infection at a given time, women with a persistent infection with high-risk HPV types were at highest risk for developing HSIL (HR = 47.1, 95% CI = 16.3 to 136, adjusting for HIV type in addition to the above-mentioned variables; Table 2 ). Women with a transient infection with high-risk HPV types (HR = 14.0, 95% CI = 3.7 to 53.5) and women with a persistent (HR = 19.1, 95% CI = 5.9 to 61.7) or transient (HR = 18.7, 95% CI = 5.9 to 59.3) infection with untyped or other HPV types were also statistically significantly more likely to develop HSIL than women who were always HPV negative. Development of HSIL was not associated with the detection or persistence low-risk HPV types (i.e., HPV 6 and HPV 11). HSIL risks estimates for transient and persistent infections with high-risk and untyped/other HPV types were similar between HIV-negative women and HIV-positive women ( Pinteraction between incident high-risk, persistent high-risk, incident untyped/other, or persistent untyped/other HPV and HIV infection = .7, .7, .9, and .9, respectively). Among HIV-infected women, HIV type (HIV-1, HIV-2, or both HIV-1 and HIV-2) was not independently associated with risk of developing HSIL after adjusting for HPV type and persistence. With regard to HPV persistence, there were no statistically significant differences in parameter estimates for risk for HSIL associated with transient and persistent infections with the various HPV types when analyses were conducted by defining HPV persistence as three consecutive visits instead of two consecutive visits.

Table 2.

Multivariable Cox regression analysis examining risk of developing high-grade squamous intraepithelial lesion (HSIL) in relation to human immunodeficiency virus (HIV) status and human papillomavirus (HPV) type and persistence *

Characteristic HR (95% CI) 
HIV status  
    HIV negative 1.0 (ref.) 
    HIV-1 only 1.0 (0.6 to 1.7) 
    HIV-2 only 0.5 (0.2 to 1.6) 
    Dual HIV-1 and HIV-2 infection 1.6 (0.5 to 4.7) 
Detection of HPV DNA  
    No HPV DNA 1.0 (ref.) 
    Transient HPV 6 or HPV 11 DNA only 0.0 (N/A) 
    Persistent HPV 6 or HPV 11 DNA only 0.0 (N/A) 
    Transient untyped/other HPV DNA only  18.7 (5.9 to 59.3) 
    Persistent untyped/other HPV DNA only  19.1 (5.9 to 61.7) 
    Transient high-risk HPV DNA  14.0 (3.7 to 53.5) 
    Persistent high-risk HPV DNA  47.1 (16.3 to 135.9) 
Characteristic HR (95% CI) 
HIV status  
    HIV negative 1.0 (ref.) 
    HIV-1 only 1.0 (0.6 to 1.7) 
    HIV-2 only 0.5 (0.2 to 1.6) 
    Dual HIV-1 and HIV-2 infection 1.6 (0.5 to 4.7) 
Detection of HPV DNA  
    No HPV DNA 1.0 (ref.) 
    Transient HPV 6 or HPV 11 DNA only 0.0 (N/A) 
    Persistent HPV 6 or HPV 11 DNA only 0.0 (N/A) 
    Transient untyped/other HPV DNA only  18.7 (5.9 to 59.3) 
    Persistent untyped/other HPV DNA only  19.1 (5.9 to 61.7) 
    Transient high-risk HPV DNA  14.0 (3.7 to 53.5) 
    Persistent high-risk HPV DNA  47.1 (16.3 to 135.9) 
*

Hazard ratios (HRs) are from multivariable Cox regression analysis, with simultaneous adjustment of all factors in the table as well as age, employment as a commercial sex worker, parity status, birth in Senegal, and cytology reader. ref. = referent; CI = confidence interval; N/A = not available.

“Untyped/other” HPV refers to HPV type 26, 40, 42, 53, 54, 55, 56, 57, 58, 59, 66, 68, 73, 82, 83, or 84 or untyped HPV, without detection of high-risk HPV type 16, 18, 31, 33, 35, 39, or 45.

High-risk HPV refers to HPV type 16, 18, 31, 33, 35, 39, or 45.

Risk of Developing HSIL in HIV-Positive Women in Relation to HIV Type and to HPV Type, Persistence, Immunosuppression, and Viral Load

To further explore the hypothesis that HIV-associated immune suppression and induction of persistent HPV infection increase the risk of progression to HSIL, we next evaluated the risk of developing HSIL among HIV-infected women in relation to persistent and transient HPV detection and to factors related to HIV-associated immunosuppression, including HIV type, CD4 count, and the level of plasma HIV RNA. Multivariable time-dependent Cox regression analyses were performed among HIV-infected women, after adjusting for cytology reader, type of Pap smear (conventional versus Thin Prep), age, employment as a commercial sex worker, parity, and birthplace in Senegal versus elsewhere ( Table 3 ). In model 1, women infected with HIV-2 were less likely to develop HSIL (HR = 0.3, 95% CI = 0.1 to 0.8) than women infected with HIV-1, whereas women dually infected with HIV-1 and HIV-2 and women infected with HIV-1 alone were at similar risk (HR = 1.4, 95% CI = 0.5 to 4.1). Risk of developing HSIL increased with decreasing CD4 counts as assessed during follow-up (model 2); women with CD4 counts between 200 and 500 cells per microliter were 2.2 times (HR = 2.2, 95% CI = 0.8 to 6.3) more likely and women with CD4 counts of less than 200 cells per microliter were 5.5 times (HR = 5.5, 95% CI = 2.0 to 15.2) more likely to develop HSIL than women with CD4 counts more than 500 cells per microliter. Among HIV-infected women, the level of HIV plasma RNA was associated with a statistically significantly increased risk of developing HSIL (HR for each order of magnitude increase in level of HIV plasma RNA = 1.4, 95% CI = 1.1 to 1.7; model 3). This increased risk for HSIL associated with increasing levels of HIV plasma RNA was similar among those infected with HIV-1 and those infected with HIV-2 ( P = .5; data not shown).

Table 3.

Risk of developing high-grade squamous intraepithelial lesions (HSILS) in human immunodeficiency virus (HIV)-positive women in relation to HIV type, level of immune suppression, plasma HIV RNA level, and human papillomavirus (HPV) type and persistence

  Model 1
 
Model 2
 
Model 3
 
Model 4
 
Model 5
 
Characteristic HR (95% CI) HR (95% CI) HR (95% CI) HR (95% CI)  HR * (95% CI)  
HIV type      
    HIV-1 only 1.0 (ref.) — — — 1.0 (ref.) 
    HIV-2 only 0.3 (0.1 to 0.9) — — — 0.7 (0.2 to 2.3) 
    Dual HIV-1 and HIV-2 infection 1.5 (0.5 to 4.5) — — — 0.4 (0.1 to 2.2) 
CD4 count      
    >500 cells per mL — 1.0 (ref.) — — 1.0 (ref.) 
    200–500 cells per mL — 2.2 (0.8 to 6.3) — — 1.6 (0.5 to 4.9) 
    <200 cells per mL — 5.5 (2.0 to 15.2) — — 1.7 (0.5 to 5.6) 
Plasma RNA level      
    HIV RNA (per order of magnitude increase) — — 1.4 (1.1 to 1.7) — 1.0 (0.8 to 1.4) 
Detection of HPV DNA      
    No HPV DNA — — — 1.0 (ref.) 1.0 (ref) 
    Transient HPV 6 or HPV 11 DNA only — — — 1.0 (ref.) 1.0 (ref) 
    Persistent HPV 6 or HPV 11 DNA only — — — 1.0 (ref.) 1.0 (ref) 
    Transient untyped/other HPV DNA only  — — — 1.0 (ref.) 1.0 (ref) 
    Persistent untyped/other HPV DNA only  — — — 5.1 (1.6 to 16.0) 4.5 (1.4 to 14.6) 
    Transient high-risk HPV DNA  — — — 3.2 (0.6 to 16.7) 2.6 (0.5 to 14.0) 
    Persistent high-risk HPV DNA  — — — 14.3 (5.6 to 36.8) 10.5 (3.7 to 30.0) 
  Model 1
 
Model 2
 
Model 3
 
Model 4
 
Model 5
 
Characteristic HR (95% CI) HR (95% CI) HR (95% CI) HR (95% CI)  HR * (95% CI)  
HIV type      
    HIV-1 only 1.0 (ref.) — — — 1.0 (ref.) 
    HIV-2 only 0.3 (0.1 to 0.9) — — — 0.7 (0.2 to 2.3) 
    Dual HIV-1 and HIV-2 infection 1.5 (0.5 to 4.5) — — — 0.4 (0.1 to 2.2) 
CD4 count      
    >500 cells per mL — 1.0 (ref.) — — 1.0 (ref.) 
    200–500 cells per mL — 2.2 (0.8 to 6.3) — — 1.6 (0.5 to 4.9) 
    <200 cells per mL — 5.5 (2.0 to 15.2) — — 1.7 (0.5 to 5.6) 
Plasma RNA level      
    HIV RNA (per order of magnitude increase) — — 1.4 (1.1 to 1.7) — 1.0 (0.8 to 1.4) 
Detection of HPV DNA      
    No HPV DNA — — — 1.0 (ref.) 1.0 (ref) 
    Transient HPV 6 or HPV 11 DNA only — — — 1.0 (ref.) 1.0 (ref) 
    Persistent HPV 6 or HPV 11 DNA only — — — 1.0 (ref.) 1.0 (ref) 
    Transient untyped/other HPV DNA only  — — — 1.0 (ref.) 1.0 (ref) 
    Persistent untyped/other HPV DNA only  — — — 5.1 (1.6 to 16.0) 4.5 (1.4 to 14.6) 
    Transient high-risk HPV DNA  — — — 3.2 (0.6 to 16.7) 2.6 (0.5 to 14.0) 
    Persistent high-risk HPV DNA  — — — 14.3 (5.6 to 36.8) 10.5 (3.7 to 30.0) 
*

Hazard ratios (HRs) are from multivariable Cox regression analysis, adjusted for age, employment as a commercial sex worker, parity status, birth in Senegal, and cytology reader. ref. = referent; — = factor not included in model.

“Untyped/other” HPV refers to HPV type 26, 40, 42, 53, 54, 55, 56, 57, 58, 59, 66, 68, 73, 82, 83, or 84 or untyped HPV, without detection of high-risk HPV type 16, 18, 31, 33, 35, 39, or 45.

High-risk HPV refers to HPV type 16, 18, 31, 33, 35, 39, or 45.

Compared with an HPV-negative status or a transient infection with untyped or other HPV, persistence of high-risk HPV types (HR = 14.3, 95% CI = 5.6 to 36.8) or persistence of untyped or other HPV types (HR = 5.1, 95% CI = 1.6 to 16.0; model 4) was associated with an increased risk of developing HSIL. Transient infection with high-risk HPV was not statistically significantly associated with increased risk for development of HSIL (HR = 3.2, 95% CI = 0.6 to 16.7). In our final multivariable analyses that modeled the potential independent effects of HIV type, CD4 counts, HIV plasma viral load, and HPV types and persistence (model 5), only persistent infection with high-risk HPV types (HR = 10.5, 95% CI = 3.7 to 30.0) and persistent infection with untyped/other HPV types (HR = 4.5, 95% CI = 1.4 to 14.6) were statistically significantly associated with risk for development of HSIL. That is, HIV type, level of CD4 immune suppression, and plasma HIV RNA level were not statistically significantly associated with independent increased risk for developing HSIL after adjustment for HPV persistence.

D ISCUSSION

In this study, we found that HIV-infected women were at greatest risk for development of HSIL and that transient and especially persistent infections with oncogenic HPV types were strongly associated with HSIL risk. HIV-positive women infected with HIV-2 were less likely to develop HSIL than HIV-positive women infected with HIV-1. HIV-positive women with CD4 + cell counts of less than 200 cells per microliter were at greater risk of HSIL than HIV-positive women with CD4 counts of more than 500 cells per microliter. High plasma HIV RNA levels were also associated with increased HSIL risk. After adjustment for HPV types and persistence, however, HIV type, plasma HIV RNA levels, and CD4 counts were no longer statistically significantly associated with an increased risk of HSIL. Thus, risk of HSIL appeared to be primarily associated with increased HPV persistence, and this increased persistence probably results from immunosuppression related to HIV-1 and/or HIV-2 infection.

This study represented a unique opportunity to assess the relative importance of HIV-1 and HIV-2 infection, HIV-associated immunosuppression, and plasma load, as well as detection and persistence of various HPV types. This longitudinal study of HIV and cervical neoplasia was conducted among participants who had never received antiretroviral therapy and had not been previously screened for cervical abnormalities. Consequently, this study was not complicated by these factors. HIV-2 is rarely detected in the United States; however, both HIV-1 and HIV-2 are common in many West African countries, India, Brazil, and Europe. HIV-2 infection is associated with a longer period of subclinical infection, lower plasma RNA loads, and a slower decline in the number of CD4 cells than HIV-1 infection. Although previous studies have evaluated associations between SIL and infection with both HIV-1 and HIV-2 ( 57 , 11 , 26 ) , none had a longitudinal study design.

In this study, we found that, among women who tested negative for high-risk HPV DNA at study entry, increased risk of development of HSIL was not statistically significantly associated with HIV infection. However, among women who tested positive for high-risk HPV DNA at study entry, infection with HIV-1 or with both HIV-1 and HIV-2, but not HIV-2 alone, was associated with an increased risk of incident HSIL compared with no HIV infection. After adjustment for the increased levels of immunosuppression, as measured by CD4 counts over time, and the increased HPV detection and persistence experienced by women infected with HIV-1 compared with women infected with HIV-2, this difference was somewhat attenuated and no longer statistically significant.

At first glance this finding seems to contradict results of our previous case–control studies in women that showed that HIV-2 infection is associated with an increased risk for invasive cervical cancer, compared with HIV-1 infection or with an HIV-negative status ( 7 ) . Given the slow decline in CD4 count that is characteristic of HIV-2 infection, women infected with HIV-2 live statistically significantly longer than those with HIV-1, especially if they do not receive antiretroviral therapy. Development of cervical cancer is a relatively slow process because, in addition to requiring a persistent HPV infection, which is associated with mild immunosuppression ( 27 ) , exposure to additional cocarcinogens and acquisition of many genetic and epigenetic changes are necessary for progression to malignancy. Furthermore, our studies ( 1 , 28 ) and those of others ( 29 ) among HIV-negative women have shown that development of HSIL is an early event. Thus, although the increased immune suppression associated with HIV-1 infection, compared with HIV-2 infection, is associated with increased risk for incident HSIL in the relatively short length of observation in this study (2–3 years), we speculate that the longer survival experienced by women infected with HIV-2 results in a greater overall lifetime risk for development of invasive cancer compared with that experienced by women infected with HIV-1. These findings could have implications for HIV-1–positive women who are receiving highly active antiretroviral therapy as well. Women receiving highly active antiretroviral therapy will likely experience increased years of mild immunosuppression with long-term alteration of HPV control, and such women may be at substantially increased risk for HPV-associated HSIL and development of invasive cervical cancer.

Previous studies ( 10 , 26 , 3032 ) have also reported that HIV infection was associated with an increased risk of detection and persistence of HPV, and multiple types of HPV. Some ( 14 , 33 ) but not other ( 12 , 13 ) previous studies have reported that HIV infection was not associated with an increased risk for development of HSIL after adjusting for HPV, although none of these studies was able to assess these factors in relationship to HIV-2 infection alone or to dual infection with HIV-1 and HIV-2. However, several studies ( 12 , 33 ) did not characterize persistence of high-risk HPV detection, which is an important predictor of development of HSIL and cervical cancer ( 3 , 4 , 23 , 3437 ) .

Few studies have been able to evaluate risk of HSIL in relation to HPV and HIV infection among women who were not receiving antiretroviral therapy. Previous studies of untreated women include that of Six et al. ( 12 ) and a multicenter study ( 15 ) in Europe that examined 229 HIV-infected women who were monitored for a median of 2 years. However, these studies contained few HSIL lesions (n = 6 in each study). In the large study by Delmas et al. ( 15 ) , HPV and HIV-associated immunosuppression was assessed at only one visit. Likewise in the ALIVE cohort ( 14 ) , HPV persistence was strongly associated with detection of SIL; in stratified analyses, repeated HPV-positive tests explained the observed association between HIV infection and SIL. However, only 11 HSILs were observed in this cohort, providing only limited power to evaluate specific risk factors for HSIL. Ellerbrock et al. ( 13 ) reported on 10 patients with incident HSIL among 328 HIV-infected and 325 HIV-uninfected women, but approximately half of the HIV-positive participants were on antiretroviral therapy. Persistent HPV infection was strongly associated with HIV infection and with development of any SIL, although surprisingly, CD4 count was not statistically significantly associated with incident SIL.

There are several possible limitations to our study. First, the study endpoint HSIL was based on cytology rather than on histology, which is generally considered the “gold standard” for classification of cervical lesions. However, cytologic endpoints have been used in many other large cohort studies of HPV and cervical disease, and recent evidence suggests that that histopathology of cervical biopsies is not more reproducible than monolayer cytology ( 38 ) . Further, although the use of cytologic rather than histologic endpoints may have led to misclassification of our outcome of interest in some cases, this potential misclassification is unlikely to have differed by HIV status ( 39 ) .

Another possible limitation is related to the definition of persistence. Studies that have examined the association between HPV persistence and the risk for cervical neoplasia development have defined persistence in many ways ( 40 ) . Prevalently detected HPV infections last an average of 7–9 months ( 23 , 24 ) and incident infections, at least in young women, remain HPV positive for 1–2 years ( 25 , 41 ) . From a practical standpoint, persistence of an HPV infection can be defined as its detection two or more times over a certain period ( 42 ) . However, there is no consensus as to the length of time that defines a persistent infection. In statistical analyses, HPV persistence has been defined in a variety of ways in prospective studies of the risk of cervical neoplasia. HPV persistence has often been defined as a HPV-positive status in consecutive visits ( 22 , 42 ) , although few longitudinal studies have used follow-up visits with only 4 months in between visits as in our study. In other studies ( 1 , 13 , 44 ) , HPV detection in multiple or most HPV-positive visits over a given period has been used to define persistence. Still others ( 45 , 46 ) have defined persistence as time from a prevalent HPV-positive status until clearing of the HPV infection (not detecting HPV at one or more visits). For ease of presentation, study subjects have usually been classified as having had persistent infection (or not) on the basis of one of the criteria above, and this classification has not been allowed to vary over time in the statistical analyses. An advantage of this study is that HPV status and persistence were evaluated at every visit, allowing a given woman to have periods of nondetection, transient HPV infection, and HPV persistence during follow-up, and the Cox regression analysis with time-dependent covariates allowed for and evaluated this variability in HPV status over time in a single woman. We defined persistence as the presence of an HPV type at two consecutive visits. However, we found no statistically significant differences in parameter estimates for risk for HSIL associated with the various transient and persistent HPV types when analyses were conducted by defining HPV persistence as three consecutive visits instead of two consecutive visits.

Another possible limitation is selection bias. As with all prospective studies, bias may have arisen because of selection of study participants and loss to follow-up, although our high participation rate and retention was high (>70% of women approached agreed to enroll and 79% of enrolled women were monitored for more than 1 year) indicates that this bias is unlikely in our study. Further, the length of follow-up was similar between women with and without HIV infection, and analyses were adjusted for the number of follow-up visits in an attempt to adjust for opportunity to assess relevant exposures and HSIL outcome.

Finally, our evaluation of development of HSIL in women infected with HIV-2 and women coinfected with HIV-1 and HIV-2, compared with women infected with HIV-1, was limited by the few women with these infections enrolled and monitored for development of incident HSIL. Improved screening methodologies and increased follow-up of HPV infection in those infected with HIV-1 or HIV-2 is needed, especially in view of the increasing availability of antiretroviral treatments.

Written informed consent was obtained according to procedures approved by the Human Subjects Committee of the University of Washington and the Senegalese National AIDS Committee.
Supported by grants from the National Cancer Institute (CA 62801 and CA 97275).
Portions of this data were presented at the 5th International AIDS Malignancy Conference, Bethesda, MD, April 2001.
We thank Deana Rich, Elise Reay-Ellers, and Macoumba Touré for their invaluable coordination of study procedures in Senegal; Mame Dieumbe Mbengue-Ly, Marie Pierre Sy, and Pierre Ndiaye for patient care; Diouana Ba and Haby Agne for their leadership in the laboratory; Giselle Woto-Gaye and Katherine Zimmerman for the reading of study Pap smears; Fatou Faye-Diop for data entry; and Alison Starling for questionnaire development and data management. We also thank Shirley Kwok, Rich Respess, and Kelly Lagassic for work in the development of assays and testing of study samples for HIV-1 and HIV-2 plasma RNA quantification; Jim Hughes and Mary Redman for help in statistical analyses; and King Holmes and Laura Koutsky for their helpful comments and discussions regarding the manuscript. Finally, we thank the women who participated in this longitudinal study.

References

(1)
Koutsky LA, Holmes KK, Critchlow CW, Stevens CE, Paavonen J, Beckmann AM, et al. A cohort study of the risk of cervical intraepithelial neoplasia grade 2 or 3 in relation to papillomavirus infection.
N Engl J Med
 
1992
;
327
:
1272
–8.
(2)
Bosch FX, Manos MM, Munoz N, Sherman M, Jansen AM, Peto J, et al. Prevalence of human papillomavirus in cervical cancer: a worldwide perspective. International biological study on cervical cancer (IBSCC) Study Group.
J Natl Cancer Inst
 
1995
;
87
:
796
–802.
(3)
Ho GY, Burk RD, Klein S, Kadish AS, Chang CJ, Palan P, et al. Persistent genital human papillomavirus infection as a risk factor for persistent cervical dysplasia.
J Natl Cancer Inst
 
1995
;
87
:
1365
–71.
(4)
Wallin KL, Wiklund F, Angstrom T, Bergman F, Stendahl U, Wadell G, et al. Type-specific persistence of human papillomavirus DNA before the development of invasive cervical cancer.
N Engl J Med
 
1999
;
341
:
1633
–8.
(5)
Seck AC, Faye MA, Critchlow CW, Mbaye AD, Kuypers J, Woto-Gaye G, et al. Cervical intraepithelial neoplasia and human papillomavirus infection among Senegalese women seropositive for HIV-1 or HIV-2 or seronegative for HIV.
Int J STD AIDS
 
1994
;
5
:
189
–93.
(6)
Langley CL, Benga-De E, Critchlow CW, Ndoye I, Mbengue-Ly MD, Kuypers J, et al. HIV-1, HIV-2, human papillomavirus infection and cervical neoplasia in high-risk African women.
AIDS
 
1996
;
10
:
413
–7.
(7)
Hawes SE, Critchlow CW, Faye Niang MA, Diouf MB, Diop A, Toure P, et al. Increased risk of high-grade cervical squamous intraepithelial lesions and invasive cervical cancer among African women with human immunodeficiency virus type 1 and 2 infections.
J Infect Dis
 
2003
;
188
:
555
–63.
(8)
Vermund SH, Kelley KF, Klein RS, Feingold AR, Schreiber K, Munk G, et al. High risk of human papillomavirus infection and cervical squamous intraepithelial lesions among women with symptomatic human immunodeficiency virus infection.
Am J Obstet Gynecol
 
1991
;
165
:
392
–400.
(9)
Wright TC Jr, Ellerbrock TV, Chiasson MA, Van Devanter N, Sun XW. Cervical intraepithelial neoplasia in women infected with human immunodeficiency virus: prevalence, risk factors, and validity of Papanicolaou smears. New York Cervical Disease Study.
Obstet Gynecol
 
1994
;
84
:
591
–7.
(10)
Miotti PG, Dallabetta GA, Daniel RW, Canner JK, Chiphangwi JD, Liomba GN, et al. Cervical abnormalities, human papillomavirus, and human immunodeficiency virus infections in women in Malawi.
J Infect Dis
 
1996
;
173
:
714
–7.
(11)
La Ruche G, Ramon R, Mensah-Ado I, Bergeron C, Diomande M, Sylla-Koko F, et al. Squamous intraepithelial lesions of the cervix, invasive cervical carcinoma, and immunosuppression induced by human immunodeficiency virus in Africa. Dyscer-CI Group.
Cancer
 
1998
;
82
:
2401
–8.
(12)
Six C, Heard I, Bergeron C, Orth G, Poveda JD, Zagury P, et al. Comparative prevalence, incidence and short-term prognosis of cervical squamous intraepithelial lesions amongst HIV-positive and HIV-negative women.
AIDS
 
1998
;
12
:
1047
–56.
(13)
Ellerbrock TV, Chiasson MA, Bush TJ, Sun XW, Sawo D, Brudney K, et al. Incidence of cervical squamous intraepithelial lesions in HIV-infected women.
JAMA
 
2000
;
283
:
1031
–7.
(14)
Ahdieh L, Munoz A, Vlahov D, Trimble CL, Timpson LA, Shah K. Cervical neoplasia and repeated positivity of human papillomavirus infection in human immunodeficiency virus-seropositive and -seronegative women.
Am J Epidemiol
 
2000
;
151
:
1148
–57.
(15)
Delmas MC, Larsen C, van Benthem B, Hamers FF, Bergeron C, Poveda JD, et al. Cervical squamous intraepithelial lesions in HIV-infected women: prevalence, incidence and regression. European Study Group on Natural History of HIV Infection in Women.
AIDS
 
2000
;
14
:
1775
–84.
(16)
Schuman P, Ohmit SE, Klein RS, Duerr A, Cu-Uvin S, Jamieson DJ, et al. Longitudinal study of cervical squamous intraepithelial lesions in human immunodeficiency virus (HIV)-seropositive and at-risk HIV-seronegative women.
J Infect Dis
 
2003
;
188
:
128
–36.
(17)
The 1988 Bethesda System for reporting cervical/vaginal cytological diagnoses. National Cancer Institute Workshop.
JAMA
 
1989
;
262
:
931
–4.
(18)
Kuypers JM, Critchlow CW, Gravitt PE, Vernon DA, Sayer JB, Manos MM, et al. Comparison of dot filter hybridization, Southern transfer hybridization, and polymerase chain reaction amplification for diagnosis of anal human papillomavirus infection.
J Clin Microbiol
 
1993
;
31
:
1003
–6.
(19)
Gravitt PE, Peyton CL, Apple RJ, Wheeler CM. Genotyping of 27 human papillomavirus types by using L1 consensus PCR products by a single-hybridization, reverse line blot detection method.
J Clin Microbiol
 
1998
;
36
:
3020
–7.
(20)
Gottlieb GS, Sow PS, Hawes SE, Ndoye I, Redman M, Coll-Seck AM, et al. Equal plasma viral loads predict a similar rate of CD4+ T cell decline in human immunodeficiency virus (HIV) type 1- and HIV-2-infected individuals from Senegal, West Africa.
J Infect Dis
 
2002
;
185
:
905
–14.
(21)
Cox DR, Oakes D. Analysis of survival data
1984
; London (UK): Chapman & Hall.
(22)
McNamee R. Confounding and confounders.
Occup Environ Med
 
2003
;
60
:
227
–34; quiz 164, 234.
(23)
Ho GY, Bierman R, Beardsley L, Chang CJ, Burk RD. Natural history of cervicovaginal papillomavirus infection in young women.
N Engl J Med
 
1998
;
338
:
423
–8.
(24)
Franco EL, Villa LL, Sobrinho JP, Prado JM, Rousseau MC, Desy M, et al. Epidemiology of acquisition and clearance of cervical human papillomavirus infection in women from a high-risk area for cervical cancer.
J Infect Dis
 
1999
;
180
:
1415
–23.
(25)
Richardson H, Kelsall G, Tellier P, Voyer H, Abrahamowicz M, Ferenczy A, et al. The natural history of type-specific human papillomavirus infections in female university students.
Cancer Epidemiol Biomarkers Prev
 
2003
;
12
:
485
–90.
(26)
Vernon SD, Unger ER, Piper MA, Severin ST, Wiktor SZ, Ghys PD, et al. HIV and human papillomavirus as independent risk factors for cervical neoplasia in women with high or low numbers of sex partners.
Sex Transm Infect
 
1999
;
75
:
258
–60.
(27)
Critchlow CW, Hawes SE, Kuypers JM, Goldbaum GM, Holmes KK, Surawicz CM, et al. Effect of HIV infection on the natural history of anal human papillomavirus infection.
AIDS
 
1998
;
12
:
1177
–84.
(28)
Krone MR, Kiviat NB, Koutsky LA. The epidemiology of cervical neoplasms. In Luesley D, Jordan J, Richart RM, editors. Intraepithelial neoplasia of the lower genital tract. Edinburgh (UK): Churchill Livingstone;
1995
. p. 49–60.
(29)
Woodman CB, Collins S, Winter H, Bailey A, Ellis J, Prior P, et al. Natural history of cervical human papillomavirus infection in young women: a longitudinal cohort study.
Lancet
 
2001
;
357
:
1831
–6.
(30)
Sun XW, Ellerbrock TV, Lungu O, Chiasson MA, Bush TJ, Wright TC Jr. Human papillomavirus infection in human immunodeficiency virus-seropositive women.
Obstet Gynecol
 
1995
;
85
:
680
–6.
(31)
Minkoff H, Feldman J, DeHovitz J, Landesman S, Burk R. A longitudinal study of human papillomavirus carriage in human immunodeficiency virus-infected and human immunodeficiency virus-uninfected women.
Am J Obstet Gynecol
 
1998
;
178
:
982
–6.
(32)
Eckert LO, Watts DH, Koutsky LA, Hawes SE, Stevens CE, Kuypers J, et al. A matched prospective study of human immunodeficiency virus serostatus, human papillomavirus DNA, and cervical lesions detected by cytology and colposcopy.
Infect Dis Obstet Gynecol
 
1999
;
7
:
158
–64.
(33)
Duerr A, Kieke B, Warren D, Shah K, Burk R, Peipert JF, et al. Human papillomavirus-associated cervical cytologic abnormalities among women with or at risk of infection with human immunodeficiency virus.
Am J Obstet Gynecol
 
2001
;
184
:
584
–90.
(34)
Remmink AJ, Walboomers JM, Helmerhorst TJ, Voorhorst FJ, Rozendaal L, Risse EK, et al. The presence of persistent high-risk HPV genotypes in dysplastic cervical lesions is associated with progressive disease: natural history up to 36 months.
Int J Cancer
 
1995
;
61
:
306
–11.
(35)
Chua KL, Hjerpe A. Persistence of human papillomavirus (HPV) infections preceding cervical carcinoma.
Cancer
 
1996
;
77
:
121
–7.
(36)
Londesborough P, Ho L, Terry G, Cuzick J, Wheeler C, Singer A. Human papillomavirus genotype as a predictor of persistence and development of high-grade lesions in women with minor cervical abnormalities.
Int J Cancer
 
1996
;
69
:
364
–8.
(37)
Schlecht NF, Kulaga S, Robitaille J, Ferreira S, Santos M, Miyamura RA, et al. Persistent human papillomavirus infection as a predictor of cervical intraepithelial neoplasia.
JAMA
 
2001
;
286
:
3106
–14.
(38)
Stoler MH, Schiffman M. Interobserver reproducibility of cervical cytologic and histologic interpretations: realistic estimates from the ASCUS-LSIL Triage Study.
JAMA
 
2001
;
285
:
1500
–5.
(39)
Branca M, Rossi E, Alderisio M, Migliore G, Morosini PL, Vecchione A, et al. Performance of cytology and colposcopy in diagnosis of cervical intraepithelial neoplasia (CIN) in HIV-positive and HIV-negative women.
Cytopathology
 
2001
;
12
:
84
–93.
(40)
Goldie SJ, Grima D, Kohli M, Wright TC, Weinstein M, Franco E. A comprehensive natural history model of HPV infection and cervical cancer to estimate the clinical impact of a prophylactic HPV-16/18 vaccine.
Int J Cancer
 
2003
;
106
:
896
–904.
(41)
Xi LF, Carter JJ, Galloway DA, Kuypers J, Hughes JP, Lee SK, et al. Acquisition and natural history of human papillomavirus type 16 variant infection among a cohort of female university students.
Cancer Epidemiol Biomarkers Prev
 
2002
;
11
:
343
–51.
(42)
Schiffman M, Kjaer SK. Chapter 2: Natural history of anogenital human papillomavirus infection and neoplasia.
J Natl Cancer Inst Monogr
 
2003
;
31
:
14
–9.
(43)
Cuschieri KS, Whitley MJ, Cubie HA. Human papillomavirus type specific DNA and RNA persistence—implications for cervical disease progression and monitoring.
J Med Virol
 
2004
;
73
:
65
–70.
(44)
Moscicki AB, Hills N, Shiboski S, Powell K, Jay N, Hanson E, et al. Risks for incident human papillomavirus infection and low-grade squamous intraepithelial lesion development in young females.
JAMA
 
2001
;
285
:
2995
–3002.
(45)
Branca M, Garbuglia AR, Benedetto A, Cappiello T, Leoncini L, Migliore G, et al. Factors predicting the persistence of genital human papillomavirus infections and PAP smear abnormality in HIV-positive and HIV-negative women during prospective follow-up.
Int J STD AIDS
 
2003
;
14
:
417
–25.
(46)
Moscicki AB, Ellenberg JH, Farhat S, Xu J. Persistence of human papillomavirus infection in HIV-infected and -uninfected adolescent girls: risk factors and differences, by phylogenetic type.
J Infect Dis
 
2004
;
190
:
37
–45.