Objective
The objective of the study was to estimate the impact of human immunodeficiency virus (HIV) infection on the incidence of high-grade cervical intraepithelial neoplasia (CIN).
Study Design
HIV-seropositive and comparison seronegative women enrolled in a prospective US cohort study were followed up with semiannual Papanicolaou testing, with colposcopy for any abnormality. Histology results were retrieved to identify CIN3+ (CIN3, adenocarcinoma in situ, and cancer) and CIN2+ (CIN2 and CIN3+). Annual detection rates were calculated and risks compared using a Cox analysis. Median follow-up (interquartile range) was 11.0 (5.4–17.2) years for HIV-seronegative and 9.9 (2.5–16.0) for HIV-seropositive women.
Results
CIN3+ was diagnosed in 139 HIV-seropositive (5%) and 19 HIV-seronegative women (2%) ( P < .0001), with CIN2+ in 316 (12%) and 34 (4%) ( P < .0001). The annual CIN3+ detection rate was 0.6 per 100 person-years in HIV-seropositive women and 0.2 per 100 person-years in seronegative women ( P < .0001). The CIN3+ detection rate fell after the first 2 years of study, from 0.9 per 100 person-years among HIV-seropositive women to 0.4 per 100 person-years during subsequent follow-up ( P < .0001). CIN2+ incidence among these women fell similarly with time, from 2.5 per 100 person-years during the first 2 years after enrollment to 0.9 per 100 person-years subsequently ( P < .0001). In Cox analyses controlling for age, the hazard ratio for HIV-seropositive women with CD4 counts less than 200/cmm compared with HIV-seronegative women was 8.1 (95% confidence interval, 4.8–13.8) for CIN3+ and 9.3 (95% confidence interval, 6.3–13.7) for CIN2+ ( P < .0001).
Conclusion
Although HIV-seropositive women have more CIN3+ than HIV-seronegative women, CIN3+ is uncommon and becomes even less frequent after the initiation of regular cervical screening.
Compared with human immunodeficiency virus (HIV)-seronegative women, HIV-seropositive women face a higher risk for coinfection by human papillomaviruses (HPV) and abnormal Papanicolaou tests. Despite this fact, cancer incidence in HIV-seropositive women receiving cervical cancer prevention measures was not significantly increased.
The reasons underlying this discrepancy are unclear. Highly oncogenic HPV types, such as 16, are minimally increased in HIV. Women in a cancer prevention program may have a high frequency of precancers that are identified and eliminated, blocking oncogenesis. However, cervical treatments for HPV-related disease were not common among women in 2 US HIV cohorts. Alternatively, HPV infections may progress rapidly to precancer and then cancer more rapidly in HIV-seropositive than seronegative women, yet the disparity may become apparent only after years of observation. Even when assessed across time, most abnormal Papanicolaou tests in HIV-seropositive women are atypical or low grade, not the high-grade results strongly correlated with precancer. Cervical precancers may be similarly infrequent or may not progress over the short observation periods of most prior studies.
The objective of this study was to describe the incidence across time of cervical precancer among HIV-seropositive and comparison of HIV-seronegative US women in a cancer prevention program.
Materials and Methods
The Women’s Interagency HIV Study (WIHS) is a US multicenter cohort study of health outcomes among HIV-seropositive women. The study also has followed up at-risk HIV-seronegative comparison women who were frequency matched for risk factors, including age, race/ethnicity, level of education, injection drug use since 1978, and total number of sexual partners since 1980. Enrollment began on Oct. 3, 1994, at 6 study consortia and over time enrolled 4068 women, including those enrolled during expansions from 2001-2002 and 2011-2012, and was designed to ensure that the cohort reflected the evolving HIV epidemic in US women.
At each site, human subjects committees reviewed and approved the study, and all participants gave written informed consent. Follow-up continues, but this analysis includes information on histological outcomes through Sept. 30, 2013.
According to study-wide protocol, single-slide conventional Papanicolaou smears were obtained every 6 months using spatula and brush for HIV-seropositive and HIV-seronegative women. Colposcopy was required by study protocol for any epithelial cytological abnormality, including atypical squamous cells of uncertain significance.
HPV testing was performed for research only and was not used in clinical management, including for atypical squamous cells of uncertain significance triage. Biopsy results were interpreted at local sites and were not centrally reviewed. Abnormal results were categorized as cervical intraepithelial neoplasia (CIN) grades 1, 2, or 3; adenocarcinoma in situ; or cancer. Unspecified high-grade dysplasia was classified with CIN3. We examined CIN3 or worse (CIN3+) and CIN2 and worse (CIN2+) in separate analyses. Cervical disease treatments were identified by self-report, supplemented by medical record abstraction when available.
To minimize confusing prevalent with incident disease, women diagnosed with CIN3+ and CIN2+ within 6 months of study enrollment were excluded. Women who had hysterectomies at baseline were excluded, and women were censored at hysterectomy during follow-up. Those without follow-up were also excluded.
Contingency tables were generated to assess baseline patient characteristics by HIV serostatus. Pearson’s χ 2 tests were used to compare baseline characteristics between HIV-seropositive and HIV-seronegative women. Wilcoxon rank-sum tests were used to compare medians. The Kaplan-Meier method was used to calculate the cumulative incidence. The incidence rates between HIV-seropositive and HIV-seronegative women were compared using Cox models with the normal approximation to the binomial distribution. And the incidence rates before 2 years and those after 2 years were compared using the bootstrapping method. All statistical tests defined significance as P < .05 using 2-sided tests.
Results
Table 1 presents the demographic characteristics at enrollment for the 3465 women at risk for CIN3+ during follow-up (900 HIV seronegative, 2565 HIV seropositive). The median age (interquartile range) for the HIV-seropositive women was 35 (30-41) years and for HIV-seronegative women was 33 (26-40) years ( P < .0001). The HIV-seropositive women were less likely to be smokers at enrollment and were more likely to have been abstinent during the 6 months before enrollment. Although the distribution of the reported lifetime number of male partners was different, with more HIV-seropositive women reporting the extremes of partner number, the median number of partners was 10 for both HIV-seropositive and HIV-seronegative women.
| Characteristic | Overall (n = 3465) | HIV– (n = 900) | HIV+ (n = 2565) | P value HIV+ vs HIV– |
|---|---|---|---|---|
| Race, n (%) | .75 | |||
| White | 489 (14) | 119 (13) | 370 (14) | |
| Hispanic | 897 (26) | 236 (26) | 661 (26) | |
| Black | 1957 (56) | 510 (57) | 1447 (56) | |
| Other | 122 (4) | 35 (4) | 87 (3) | |
| Smoking status, n (%) | .01 | |||
| Never smoked | 1158 (34) | 278 (31) | 880 (34) | |
| Former smoker | 514 (15) | 119 (13) | 395 (15) | |
| Current smoker | 1778 (52) | 500 (56) | 1278 (50) | |
| Education, n (%) | .14 | |||
| Less than high school | 1288 (37) | 310 (35) | 978 (38) | |
| High school | 1034 (30) | 272 (30) | 762 (30) | |
| Beyond high school | 1134 (33) | 312 (35) | 822 (32) | |
| Lifetime male sexual partners, n (%) | .001 | |||
| <5 | 769 (23) | 167 (19) | 602 (24) | |
| 5-9 | 716 (21) | 194 (22) | 522 (21) | |
| 10-49 | 1142 (34) | 340 (38) | 802 (32) | |
| ≥50 | 779 (23) | 189 (21) | 590 (23) | |
| Male sexual partner in past 6 months, n (%) | < .0001 | |||
| 0 | 865 (26) | 152 (17) | 713 (2) | |
| 1 | 1833 (54) | 439 (49) | 1394 (56) | |
| 2 | 372 (11) | 146 (16) | 226 (9) | |
| ≥3 | 317 (9) | 158 (18) | 159 (6) | |
| CD4+ cell count/cmm, n (%) | ||||
| >500 | 889 (36) | |||
| 200-500 | 1099 (44) | |||
| <200 | 505 (20) | |||
| HIV viral load (copies/cmm), n (%) | ||||
| ≤4000 | 1057 (42) | |||
| 4001-20,000 | 518 (21) | |||
| 20,001-100,000 | 532 (21) | |||
| >100,000 | 403 (16) | |||
| Ever AIDS, n (%) | ||||
| No | 2037 (79) | |||
| Yes | 528 (21) |
HIV-related disease characteristics among seropositive women are summarized in Table 1 . Most HIV-seropositive women had CD4 cell counts greater than 200/cmm, with HIV RNA levels below or just above the threshold for detection at study initiation, which for most women was less than 4,000 copies/cmm. Most had never been diagnosed with acquired immunodeficiency syndrome.
Median follow-up (interquartile range) was 11.0 (5.4–17.2) years for HIV-seronegative and 9.9 (2.5–16.0) for HIV-seropositive women. Follow-up rates for HIV-seropositive and HIV-seronegative women at 3 years were 86% and 80%, 82% and 74% at 5 years, and 71% and 62% at 10 years. Only 19 HIV-seronegative women (2%) were diagnosed with CIN3+, whereas CIN3+ was found in 139 HIV-seropositive women (5%) ( P < .0001).
The risk of CIN2+ was substantially greater, occurring in 34 HIV-seronegative women (4%) and 316 HIV-seropositive women (12%) ( P < .0001). The annual detection rate of CIN3+ was 0.6 per 100 person-years in HIV-seropositive women and 0.2 per 100 person years in HIV-seronegative women ( P < .0001). Similar rates for CIN2+ were 1.4 and 0.4 per 100 person-years ( P < .0001).
The annual detection rate of CIN3+ and CIN2+ fell over time ( Figure ). We estimated the detection rate of these endpoints before and after the first 2 years after enrollment, excluding presumed prevalent disease discovered during the first 6 months of study. The annual rate of CIN3+ detection among HIV-seropositive women was 0.9 per 100 person-years during the first 2 years of study and 0.4 per 100 person-years during subsequent follow-up ( P < .0001). CIN2+ incidence among these women fell similarly with time, from 2.5 per 100 person-years during the first 2 years after enrollment to 0.9 per 100 person-years subsequently ( P < .001).
The incidence was lower among HIV-seronegative women but also fell with time: CIN3+ incidence dropped from 0.4 per 100 person-years during the first 2 years to 0.1/100 person-years thereafter ( P = .02), whereas CIN2+ incidence decreased from 0.7 per 100 person-years early in the study to 0.2 per 100 person-years afterward ( P = .03). The incidence of CIN3+ and CIN2+ was higher among HIV-seropositive than HIV-seronegative women at both time points ( P < .0001), except that the difference in CIN3+ during the first years was not significant ( P = .07). The decline in risk after 2 years persisted among HIV-seropositive women after controlling for CD4 count women and for age (hazard risk, 0.18 after 2 years; 95% confidence interval [CI], 0.08–0.40; P < .0001). There were insufficient data to conduct a similar multivariate analysis among HIV-seronegative women.
If the observed increased risk of CIN3+ and CIN2+ in HIV-seropositive women is related to immunosuppression, then risk should rise with more severe immunosuppression. In fact, the incidence of CIN3+ was 0.2 per 100 person-years (95% CI, 0.1–0.3) among HIV-seronegative women, 0.5 (95% CI, 0.3–0.6; P = .003 compared with HIV-seronegative women) among HIV-seropositive women with CD4 counts greater than 500/cmm, 0.5 (95% CI, 0.4–0.7; P = .0003) among women with CD4 counts of 200-500/cmm, and 1.0 (95% CI, 0.7–1.3; P < .0001) among women with CD4 counts less than 200/cmm.
CIN2+ risk rose similarly, from 0.4 per 100 person-years (95% CI, 0.2–0.5) among HIV-seronegative women to 1.2 (95% CI, 0.9–1.4; P < .0001 compared with HIV-seronegative women) among HIV-seropositive women with CD4 counts greater than 500/cmm, 1.4 (95% CI, 1.1–1.6; P < .0001) for women with CD4 counts of 200-500/cmm, and 2.3 (95% CI, 1.7–2.8; P < .0001) among women with CD4 counts less than 200/cmm.
Although significant, these results may obscure the impact of CD4 count on CIN3+ and CIN2+ because levels may vary over time. In an analysis adjusting for CD4 count as a time-varying factor, we observed a progressive rise in CIN3+ and CIN2+ risk as CD4 counts fell ( Table 2 ). In Cox analyses controlling for age and the timing of diagnosis, the hazard ratio for HIV-seropositive women with CD4 counts less than 200/cmm compared with HIV-seronegative women was 4.6 (95% CI, 3.0–7.2) for CIN3+ and 3.5 (95% CI, 2.6–4.7) for CIN2+ ( P < .0001 for both).
| Variable | Hazard ratio | 95% LCL | 95% UCL | P value |
|---|---|---|---|---|
| CIN3+ | ||||
| Timing of diagnosis | ||||
| Initial 2 y (referent) | 1.0 | |||
| After 2 y | 0.18 | 0.08 | 0.40 | < .0001 |
| CD4+ T-cell count | ||||
| >500 (referent) | 1.0 | |||
| 200-500 | 1.40 | 0.88 | 2.21 | .16 |
| <200 | 4.62 | 2.95 | 7.22 | < .0001 |
| Age, y | ||||
| <30 (referent) | 1.0 | |||
| 30-34 | 0.87 | 0.46 | 1.65 | .66 |
| 35-39 | 0.94 | 0.51 | 1.71 | .84 |
| 40-44 | 0.66 | 0.34 | 1.26 | .21 |
| ≥45 | 0.73 | 0.38 | 1.39 | .33 |
| CIN2+ | ||||
| Timing of diagnosis | ||||
| Initial 2 y (referent) | 1.0 | |||
| After 2 y | 0.30 | 0.17 | 0.51 | < .0001 |
| CD4+ T-cell count/cmm | ||||
| >500 (referent) | 1.0 | |||
| 200-500 | 1.52 | 1.14 | 2.03 | .004 |
| <200 | 3.45 | 2.56 | 4.65 | < .0001 |
| Age, y | ||||
| <30 (referent) | 1.0 | |||
| 30-34 | 0.60 | 0.41 | 0.87 | .01 |
| 35-39 | 0.69 | 0.49 | 0.98 | .04 |
| 40-44 | 0.39 | 0.26 | 0.59 | < .0001 |
| ≥45 | 0.46 | 0.31 | 0.69 | .0001 |
We further explored factors associated with the incidence of CIN3+ and CIN2+. As shown in Table 3 , CD4 count remained strongly associated with both CIN3+ and CIN2+. Smoking was associated with a higher and increasing age with a lower risk. Lifetime number of sexual partners was not associated with either CIN3+ or CIN2+ after controlling for these factors. Recent number of sexual partners was linked to CIN2+ but not CIN3+.
| Variable | Hazard Ratio | 95% LCL | 95% UCL | P value |
|---|---|---|---|---|
| CIN3+ | ||||
| CD4+ T-cell count | ||||
| HIV negative (referent) | 1.0 | < .0001 a | ||
| >500 | 1.88 | 1.04 | 3.40 | .04 |
| 200-500 | 2.45 | 1.40 | 4.27 | .002 |
| <200 | 7.78 | 4.51 | 13.42 | < .0001 |
| Age, y | ||||
| <30 (referent) | 1.0 | |||
| 30-34 | 0.66 | 0.37 | 1.17 | .15 |
| 35-39 | 0.64 | 0.37 | 1.10 | .11 |
| 40-44 | 0.46 | 0.25 | 0.83 | .01 |
| ≥45 | 0.53 | 0.29 | 0.97 | .04 |
| Smoking | ||||
| Never smoked (referent) | 1.0 | |||
| Former smoker | 0.85 | 0.47 | 1.54 | .60 |
| Current smoker | 2.08 | 1.35 | 3.21 | .001 |
| Lifetime male sexual partners, n | ||||
| <5 (referent) | 1.0 | |||
| 5-9 | 1.71 | 1.03 | 2.84 | .04 |
| 10-49 | 1.11 | 0.67 | 1.85 | .69 |
| ≥50 | 1.24 | 0.72 | 2.13 | .43 |
| Male sexual partners in past 6 mo, n | ||||
| 0 (referent) | 1.0 | |||
| 1 | 1.20 | 0.83 | 1.75 | .34 |
| 2 | 0.68 | 0.30 | 1.54 | .35 |
| ≥3 | 0.62 | 0.22 | 1.78 | .37 |
| CIN2+ | ||||
| CD4+ T-cell count | ||||
| HIV negative (referent) | 1.0 | < .0001 a | ||
| >500 | 2.86 | 1.89 | 4.33 | < .0001 |
| 200-500 | 4.23 | 2.86 | 6.25 | < .0001 |
| <200 | 9.50 | 6.36 | 14.18 | < .0001 |
| Age, y | ||||
| <30 (referent) | 1.0 | |||
| 30-34 | 0.53 | 0.37 | 0.76 | .001 |
| 35-39 | 0.62 | 0.45 | 0.86 | .004 |
| 40-44 | 0.33 | 0.22 | 0.48 | < .0001 |
| ≥45 | 0.44 | 0.30 | 0.65 | < .0001 |
| Smoking | ||||
| Never smoked (referent) | 1.0 | |||
| Former smoker | 0.96 | 0.66 | 1.39 | .83 |
| Current smoker | 1.87 | 1.41 | 2.47 | < .0001 |
| Lifetime male sexual partners, n | ||||
| <5 (referent) | 1.0 | |||
| 5-9 | 1.13 | 0.81 | 1.57 | .49 |
| 10-49 | 1.02 | 0.74 | 1.40 | .92 |
| ≥50 | 0.98 | 0.69 | 1.39 | .90 |
| Male sexual partners in past 6 mo, n | ||||
| 0 (referent) | 1.0 | |||
| 1 | 1.56 | 1.19 | 2.05 | .002 |
| 2 | 1.69 | 1.09 | 2.60 | .02 |
| ≥3 | 1.46 | 0.84 | 2.53 | .18 |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
