HIV-positive women have higher risk of human papilloma virus infection, precancerous lesions, and cervical cancer

Abstract

Objective: HIV-positive women have higher human papillomavirus (HPV) prevalence and cervical cancer incidence than HIV-negative women, partly because of HIV's modifying effect on HPV pathogenesis. We synthesized the literature on the impact of HIV on HPV natural history.

Design: Systematic review and meta-analysis.

Methods: We searched the literature for studies evaluating HPV acquisition and persistence or precancer progression by HIV status. Data on HPV natural history by HIV status, CD4 cell counts, viral load, and antiretroviral therapy (ART) were summarized using fixed effect models.

Results: Overall, 38 of 1845 abstracts identified met inclusion criteria. HIV-positive women had higher HPV acquisition [relative risk (RRpooled) 2.64, 95% confidence interval (CI) 2.04-3.42] and lower HPV clearance (hazard ratiopooled 0.72, 95% CI 0.62-0.84) than HIV-negative women. HPV acquisition was higher with declining CD4 cell count and was lower in those virally suppressed on ART. HIV was associated with higher incidence of low-grade squamous intraepithelial lesions (LSIL; RRpooled 3.73, 95% CI 2.62-5.32) and high-grade squamous intraepithelial lesions (HSIL; hazard ratiopooled 1.32, 95% CI 1.10-1.58), largely because of increased HPV persistence. ART lowered progression from normal cytology to LSIL (hazard ratiopooled 0.65, 95% CI 0.52-0.82), but not HSIL. Cervical cancer incidence was associated with HIV positivity (RR 4.1, 95% CI 2.3-6.6), but not with ART.

Conclusion: HIV-positive women have higher risk of acquiring HPV, with risk inversely associated with CD4 cell count. ART lowered HPV acquisition, increased clearance, and reduced precancer progression, likely via immune reconstitution. Although some of our results are limited by small number of studies, our study can inform screening guidelines and mathematical modeling for cervical cancer prevention.

Figure 1

Flow chart of the systematic review process.

Figure 2. HPV incidence among HIV-positive women compared to HIV-negative women

N, sample size; ES, effect size. A). Incidence of HPV infection by HPV type. Banura, 2008 was not included in calculating the overall effect because it reported odds ratio; other studies reported relative risk. B). Relative incidence of any HPV infection, by CD4 count. The effect estimates from Strickler et al, 2005 were stratified HIV RNA load; U = undetectable (<4,000 copies/mL), L = low (4,000–20,000 copies/mL), M = moderate (20,001–100,000 copies/mL), and H = high (>100000 copies/mL). Effect size was measured in odds ratios in Phelan et al, 2009 and hazard ratios in Xie et al, 2003 and Strickler et al, 2005. Effect estimates from Mbulawa et al 2012, Denny et al, 2008, and Mane et al, 2016 were not included in the figure. Mbulawa et al, 2012 used a different CD4 cutoff (>350 vs. <350) and Denny et al, 2008 and Mane et al, 2016 evaluated CD4 as a continuous variable.

Figure 3. Clearance of newly detected and prevalent HPV infections by HIV status and CD4 count

N: sample size, ES: effect size. N= “-” Indicates that the study did not report sample size associated with their estimates. A). Clearance of newly detected HPV infection among HIV-positive women compared to HIV-negative women. B). Clearance of prevalent and newly detected HPV infection among HIV-positive women compared to HIV-negative women, by HPV type. C). Clearance of HPV infection among HIV-positive women by CD4 count. Estimates from Mane et al, 2016 and Ahdieh et al, 2000 were not included in the forest plot. Mane et al, 2016 evaluated CD4 count as a continuous variable. And Ahdieh et al, 2000 used HIV-negative women was their reference group.

Figure 3. Clearance of newly detected and prevalent HPV infections by HIV status and CD4 count

N: sample size, ES: effect size. N= “-” Indicates that the study did not report sample size associated with their estimates. A). Clearance of newly detected HPV infection among HIV-positive women compared to HIV-negative women. B). Clearance of prevalent and newly detected HPV infection among HIV-positive women compared to HIV-negative women, by HPV type. C). Clearance of HPV infection among HIV-positive women by CD4 count. Estimates from Mane et al, 2016 and Ahdieh et al, 2000 were not included in the forest plot. Mane et al, 2016 evaluated CD4 count as a continuous variable. And Ahdieh et al, 2000 used HIV-negative women was their reference group.

Figure 4. LSIL incidence by HIV status and ART use

N: Sample size, ES: effect size. A). LSIL incidence among HIV-positive women compared to HIV-negative women. Duerr et al, 2006 reported its effect size in hazard ratios, and was not included in calculating the overall estimate. B). LSIL incidence among HIV-positive women on ART compared to HIV-positive women not on ART. Estimates from Adler et al, 2012, Firnhaber et al, 2012, and Ellerbrock et al, 2000 adjusted for CD4 count. C). LSIL incidence by absolute increase in CD4 count after ART initiation. Data extracted from Kim, 2013.

Figure 5. LSIL regression by HIV status, CD4 count, and ART use

N: Sample size, ES: effect estimate. N= “-” Indicates that the study did not report sample size associated with their estimates. A). LSIL regression among HIV-positive women compared to HIV-negative women. B). LSIL regression among HIV-positive women compared to HIV-negative women, by CD4 count. Data extracted from Six et al, 1998. Confidence intervals for these estimates were not included in the study. C). LSIL regression among HIV-positive women on ART compared to HIV-positive women not on ART. Data from Ahdieh-Grant et al, 2004 was not included in the figure. Because LSIL regression was not observed in any of the HIV-positive women not on ART, the study could not obtain an estimate for the effect of ART use on LSIL regression.