Article Text
Abstract
Objective HIV infection increases the risk of high-grade cervical neoplasia and invasive cervical carcinoma. The study addresses the limited data describing human papillomavirus (HPV) infection and cervical neoplasia among HIV-infected women in HIV-discordant relationships in sub-Saharan Africa, which is needed to inform screening strategies.
Methods A cross-sectional study of HIV-infected women with HIV-uninfected partners was conducted to determine the distribution of type-specific HPV infection and cervical cytology. This study was nested in a prospective cohort recruited between September 2007 and December 2009 in Nairobi, Kenya. Cervical cells for HPV DNA testing and conventional cervical cytology were collected. HPV types were detected and genotyped by Roche Linear Array PCR assay.
Results Among 283 women, the overall HPV prevalence was 62%, and 132 (47%) had ≥1 high-risk (HR)-HPV genotype. Of 268 women with cervical cytology results, 18 (7%) had high-grade cervical lesions or more severe by cytology, of whom 16 (89%) were HR-HPV-positive compared with 82 (41%) of 199 women with normal cytology (p<0.001). The most common HR-HPV types in women with a high-grade lesion or more severe by cytology were HPV-52 (44%), HPV-31 (22%), HPV-35 (22%), HPV-51 (22%) and HPV-58 (22%). HR-HPV genotypes HPV-16 or HPV-18 were found in 17% of women with high-grade lesions or more severe. HR-HPV screening applied in this population would detect 89% of those with a high-grade lesion or more severe, while 44% of women with normal or low-grade cytology would screen positive.
Conclusion HR-HPV prevalence was high in this population of HIV-infected women with an uninfected partner. Choice of screening for all HR genotypes versus a subset of HR genotypes in these HIV-infected women will strongly affect the performance of an HPV screening strategy relative to cytological screening. Regional and subpopulation differences in HR-HPV genotype distributions could affect screening test performance.
- HPV
- women
- cervical cytology
- cervical neoplasia
- HIV
Statistics from Altmetric.com
Introduction
Human papillomavirus (HPV) infection is the most common STI in women worldwide.1 More than 80% of sexually active adults will be infected with HPV during their lifetime, and most HPV infections present subclinically and are cleared spontaneously.1 Fourteen HPV types that infect the genital mucosa are classified as high-risk (HR) types: HPV 16, 18, 31, 33, 35, 39, 45, 51, 52, 56, 58, 59, 66 and 68.2 HR-HPV types are the primary cause of invasive cervical cancer (ICC) and are detected in up to 99.7% of ICC cases.3
Women with HIV are at increased risk of progressing from HPV infection to high-grade cervical precancerous lesions,4 and have an accelerated course from these high-grade lesions to carcinoma.5 6 ICC is the most common cancer in East African women7 and is one of the criteria in the HIV/AIDS WHO clinical stage 4 classification.8
In Kenya, the coverage of cervical cancer screening is low, with only 8% of women aged 25–45 reporting ever having been screened.9 The elevated risk of cervical abnormalities and subsequent progression to cervical cancer among HIV-infected women make them an important target for intervention. However, data are limited on HPV and cervical precancer in HIV-infected women.10 11
In this study, we sought to determine the prevalence and correlates of HPV infection and cytological cervical abnormalities, describe the distribution of HPV genotypes among HIV-infected women in HIV-discordant relationships in Nairobi, Kenya, and inform the design of screening and treatment programmes.
Methods
Study design and data collection
We conducted a cross-sectional study of HIV-infected women aged 18–50 years, nested within a prospective study of HIV-discordant couples in Nairobi, Kenya from September 2007 to December 2009, as previously described.12 Couples were recruited from ≥50 voluntary counselling and testing sites around Nairobi. Heterosexual HIV-discordant couples were eligible if they reported having sex ≥3 times in the 3 months prior to screening, were not pregnant and planned to remain together for the duration of the study. At enrolment, HIV-infected participants did not have a history of clinical AIDS (WHO stage IV) and were not currently on antiretroviral therapy (ART). Couples were ineligible if the HIV-uninfected partner reported sexual relationships outside the primary partnership. At enrolment, interviews were conducted separately to collect individual sociodemographic and sexual history data from women and their male partners.
Participants were tested for HIV-1 by two rapid tests conducted in parallel using the Determine HIV-1/2 rapid test (Abbott Laboratories, Tokyo, Japan; now marketed by Inverness Medical as Alere Determine) and the Bioline HIV 1/2 rapid test (Standard Diagnostics, Suwon, South Korea) and were eligible only if they had concordant rapid test results at screening. Blood was drawn to quantify CD4 T cell counts (BD FACSCalibur, Becton-Dickinson, Franklin Lakes, New Jersey) and plasma HIV RNA viral load (Gen-Probe transcription-mediated amplification assay).13
Cervical cytology and HPV detection
Study physicians and clinical officers performed pelvic examinations to collect exfoliated cervical cells using endocervical brushes for diagnosis of cervical abnormalities by conventional cervical cytology. An expert cytologist at the University of Nairobi Pathology Laboratory read the cytology slides. All abnormal cytology and 10% of randomly sampled normal cytology slides were blindly double-read by an independent cytologist for quality assurance. Cervical cytology was categorised according to the 2001 Bethesda system.14 Women were classified as having normal cytology if they did not have cervical intraepithelial lesions. Low-grade cervical abnormalities included atypical squamous cells of undetermined significance (ASC-US) and low-grade squamous intraepithelial lesions (LSIL). High-grade cervical abnormalities or more severe included atypical squamous cells, cannot exclude a high-grade squamous intraepithelial lesion (ASC-H) and high-grade squamous intraepithelial lesions (HSIL), and women with squamous cell carcinoma (SCC). Kenyan guidelines were available for visual inspection with acetic acid/visual inspection with Lugol’s iodine screening only, but not for conventional cervical cytology screening. We followed the standard of care at Kenyatta National Hospital, whereby women with ASC-US or LSIL were recalled for a repeat cervical cytology screening after 6 months and women with ASC-H and HSIL were scheduled for a colposcopy and management at Kenyatta National Hospital.12
During the genital examination, cervicovaginal secretions were collected by rotating a cotton swab 360° in the cervical os and rotating another swab across the vaginal wall. Both swabs were placed in the same tube containing 5 mL sterile phosphate buffered saline and immediately transported to the laboratory on ice. An aliquot was used for HPV DNA detection using Linear Array assay (Roche Molecular Systems, Pleasanton, California), according to the manufacturer’s instructions.15 The HPV Linear Array assay uses the PGMY09/11 primer system for high-efficiency amplification of 37 HPV genotypes.16 HR-HPV for this analysis included genotypes 16, 18, 31, 33, 35, 39, 45, 51, 52, 56, 58, 59, 66 and 68.2 HR-HPV infections were defined as infection with any of the HR-HPV genotypes, with or without a low-risk HPV (LR-HPV) infection. LR-HPV positivity was defined as infection with only LR-HPV genotypes (genotypes 6, 11, 26, 40, 42, 53, 54, 55, 61, 62, 64, 67, 69, 70, 71, 72, 73, 81, 82, 83, 84, 87 and 89).
Data analyses
Descriptive analysis of HPV genotype distribution was conducted stratified by cytological status of HIV-infected women. We calculated HPV prevalence ratios (PRs) and 95% CIs for potential female and male partner risk factors for HPV infection and for abnormal cervical cytology. We chose to calculate PRs as opposed to ORs because the prevalence of the outcomes was not rare, making the interpretation of the ORs more difficult. P values were based on exact χ2 tests. We investigated demographic (age, education, employment status, and marital status), biological (CD4 T cell count, plasma viral load, herpes simplex virus serology, parity, and male partner circumcision), and behavioural factors (number of sexual acts in the last month, practising dry sex, unprotected sex, and partner’s previous STIs) as risk factors for HPV infection. CD4 T cell counts were categorised as <0.200, 0.200–0.349, 0.350–0.499 or ≥0.500 x 109 cells/L based on historical thresholds for ART initiation. ASC-US and LSIL were analysed as low-grade cervical abnormalities, and ASC-H, HSIL and SCC as high-grade abnormalities or more severe. Stata V.14.0 was used for data analysis.
Results
Study participants
Of 305 HIV-infected women enrolled in the study, 283 had a valid HPV test, and of these 268 had available cytology results. The median age of women was 28 years (range 18–50), the median age of sexual debut was 17 years of age, and the median number of lifetime sex partners was 3 (table 1). A relatively small proportion of women were using hormonal contraception (20%). The median CD4 T cell count was 0.455 x 109 cells/L (IQR: 0.299–0.639) and 90 (32%) women had a CD4 T cell count <0.350 x 109 cells/L.
Cervical HPV infection
Among the 283 HIV-infected women, any HPV was detected in 176 (62%), among whom 116 (66%) had ≥2 concurrent HPV infections, with a median of 2 infections (IQR: 1–4). HR-HPV genotypes were identified in 132 (47%) women overall, of whom 101 (77%) occurred in combination with ≥1 other HPV genotype and 76 (58%) included ≥2 concurrent HR-HPV infections. Compared with women without HPV detected or with only LR-HPV, the presence of an HR-HPV genotype was associated with self-reported unprotected sex in the past month (PR=1.65; 95% CI 1.31 to 2.09; p<0.001), higher HIV-1 plasma viral load (PR=1.32 per log10 RNA copies/mL; 95% CI 1.12 to 1.56; p=0.001), and lower CD4 T cell counts (<0.500 x 109 cells/L) (PR=1.59; 95% CI 1.20 to 2.10; p<0.001) (table 1).
HPV infection in relation to cervical cytology
Of the 268 women with available cervical cytology results, 199 (74%) had normal cytology, 51 (19%) had a low-grade lesion (ASC-US or LSIL), 16 (6%) had a high-grade lesion (ASC-H or HSIL) and 2 (0.8%) had SCC. We found no sociodemographic or behavioural factors significantly associated with either low-grade or high-grade lesions (online supplementary table 1). For subsequent analyses, women with SCC were grouped with those with high-grade cervical lesions.
Supplemental material
Among women with normal cervical cytology, 118 (59%) had an HPV infection identified, of whom 82 (69%) were infected with ≥1 HR-HPV genotype and 36 (31%) were infected only with LR-HPV. The most common HPV genotype among women with normal cytology was HPV-52, which was found in 16% of women, followed by HPV-53 (12%) and HPV-68 (8%) (table 2). By comparison, among 51 women with low-grade lesions, 35 (69%) had an HPV infection, of whom 28 (80%) had HR-HPV genotypes and 7 (20%) had only LR-HPV genotypes. The most common HPV genotypes in women with low-grade lesions were HPV-52 (29%), HPV-35 (20%), HPV-18 (14%), HPV-51 (14%) and HPV-66 (14%), all of which are HR-HPV genotypes. Among 18 women with high-grade lesions or SCC, 16 (89%) had an HPV infection identified, all of whom had an HR-HPV genotype. The most common HPV genotypes were HR types HPV-52 (44%), HPV-31 (22%), HPV-35 (22%), HPV-51 (22%) and HPV-58 (22%), and LR types HPV-73 (28%) and HPV-55 (17%). Compared to women with normal cytology, HR-HPV genotypes 31, 35, 51, 52 and 58 were significantly more prevalent among women with high-grade lesions or SCC.
Single-type HR-HPV infections (with or without another LR genotype detected) were found in 7 (39%) women with high-grade lesions or SCC, among whom HPV-51 was found in 2 (33%) women, and HPV-16, HPV-18, HPV-31, HPV-39, and HPV-58 were found in 1 (17%) woman each. Compared with women with normal cytology, there was a marginally significant 1.98-fold higher prevalence (95% CI 1.04 to 3.78; p=0.07) of a single HR-HPV genotype and a significant 2.31-fold higher prevalence (95% CI 1.36 to 3.94; p=0.02) of multiple HR-HPV genotypes among HIV-infected women with high-grade lesions.
HR-HPV genotypes HPV-16 and/or HPV-18 were found in 17% of women with high-grade lesions or SCC. In comparison, HPV-16 was found in 7% of women with normal cervical cytology and 4% of those with low-grade lesions, and HPV-18 was found in 5% of women with normal cervical cytology and 14% of those with low-grade lesions. Neither type alone (p=0.99 for HPV-16 and p=0.26 for HPV-18) or in combination (p=0.44) was statistically associated with high-grade lesions or SCC.
We compared the screening performance of HPV detection versus cervical cytology to determine possible implications of screening by HPV testing in this population. A screening strategy based on detection of any HR-HPV infection would have resulted in 47% of women screening positive (table 3). Of those women with high-grade lesion or SCC by cervical cytology, 89% would have been HR-HPV-positive and 11% would have had a negative HR-HPV test. An alternative screening strategy based on detection of only HR-HPV genotypes 16, 18, 31 or 45 would have resulted in 19% of women screening positive. In this scenario, 44% of high-grade lesions or more severe would have been positive for HPV 16, 18, 31 or 45, and screening for this subset of HR-HPV genotypes would have missed 56% of the women with high-grade cytology or SCC.
Discussion
In 268 HIV-seropositive Kenyan women with discordant HIV-negative male partners, we observed a high prevalence of HPV (62%), HR-HPV genotypes (47%) and high-grade cervical cytology (6%). The most common HR-HPV infections among those with high-grade lesions or more severe were genotypes 52, 31, 35, 51, and 58. In a population of HIV-infected women under similar conditions to this study, screening for HR-HPV would detect 89% of high-grade or SCC cases and would result in positive tests in 44% of women with normal or low-grade cervical cytology. This information should be combined with cost data and mathematical models to design HPV screening, evaluation, and treatment programmes that target HIV-infected women to reduce the burden of disease from cervical cancer.
We found HPV (any genotype) in 59% of women with normal cytology, 69% of women with low-grade lesions and 94% of women with high-grade lesions, estimates generally consistent with previous findings from HIV-infected women from sub-Saharan Africa.17 18 Our observed HPV prevalence of 59% among women with normal cervical cytology was considerably higher than a global meta-analysis estimate of 36% among HIV-infected women with no cytological abnormalities, but was in line with meta-analysis estimates among HIV-infected women with normal cervical cytology from sub-Saharan Africa (57%) and South America (57%).17 18 Our findings are also consistent with another Kenyan study that found that 61% of HIV-seropositive women with normal cytology had an HPV infection of any type and 53% of HIV-seropositive women overall were infected with an HR-HPV genotype.19 This study was limited to women not currently on ART, as such these findings may not be generalisable to women on ART.
We found that 47% of HIV-infected women from HIV-discordant couples were infected with at least one HR-HPV genotype. Among all HIV-infected women, the most common HR-HPV genotype was HPV-52 (20%), followed by HPV-51 (9%) and HPV-35 (8%), and the most common LR-HPV genotype was HPV-53 (11%). The genotype distribution in our study is more consistent with other studies among HIV-seropositive women from sub-Saharan Africa than with those from other regions.20 21 In a meta-analysis examining HPV genotype distributions among HIV-infected women with normal cervical cytology in sub-Saharan Africa, HPV-16 and HPV-58 were the most prevalent HR-HPV genotypes, while HPV-53 was also the most prevalent non-HR genotype, and HPV-52 was among the top four most common HR genotypes.18 Other Kenyan studies found HPV-52 was the most common HPV type among family planning clinic attendees20 and among female sex workers.21 Similar to other findings from Western Kenya, compared with women with normal cytology, those with high-grade lesions or more severe had a twofold higher prevalence of a single HR-HPV genotype and 2.3-fold higher prevalence of multiple HR-HPV genotypes.21 22 While the cross-sectional nature of this study prevents causal conclusions, these observations provide some support to the hypothesis that, at least among HIV-infected women, multiple HR-HPV infections have a cumulative effect on the risk of cervical abnormalities or may represent an immune marker associated with an increased risk of cervical cancer.
Globally, HPV-16 and HPV-18 are the most common HPV genotypes detected in women with ICC, with meta-analyses estimates that 67%–70% of cervical carcinomas globally are attributable to these two HPV genotypes.23 24 This pattern is similar in Africa, where meta-analysis estimates indicate HPV-16/18 are detected in ~73% of women with ICC, regardless of HIV status.25 Likewise, a previous study from Nairobi found HPV-16/18 in 65% of HIV seropositive and 60% of seronegative women with ICC.26 Due to the absence of ICC cases in this study, we were unable to evaluate the contribution of HPV-16/18 specifically to ICC, but found that HPV-16 and HPV-18 were uncommon in women with high-grade lesions who did not have ICC, with each found in only 6% of women with high-grade lesions. This was even lower than a previous Kenyan study that found HPV-16 in 36% of women and HPV-18 in 4% of women attending a family planning clinic who had an HSIL cytological finding.20 It is unclear whether the lower prevalence of HPV-16 among women with high-grade lesions, and to a lesser degree HPV-18, is due to our relatively small sample size of high-grade or more severe cases (n=18), random chance, regional difference, the influence of HIV infection or dynamics related to being in an HIV-discordant relationship.
Based on our findings, HPV screening for all HR-HPV genotypes would identify 89% of high-grade lesions/SCC detected by cytology, but would also be positive for 44% of HIV-infected women with normal cytology/low-grade lesions who would have required no or less intensive follow-up based on cytology screening. Screening for all HR-HPV genotypes would be more sensitive than cytological screening, yet would likely result in identification of a large number of women who would not have been identified by cervical cytology. Conversely, screening for a smaller subset of genotypes (eg, genotypes 16, 18, 31 and 45) appears to fail to identify a relatively large proportion of women who would have been identified with high-grade lesions by cytology. An important limitation in the interpretation of these findings is that we did not have histological confirmation of the cytology results. Given the relatively low sensitivity of cytology to detect high-grade lesions, it is possible that HR-HPV screening may identify true cases of cervical lesions that were not detected by cytology.
The success of an HR-HPV screening programme will require an accurate, rapid, inexpensive and highly acceptable testing modality. We conducted a previous study of women’s knowledge of HPV and Pap smears in this population and found 82% of women were amenable to self-collection for HPV testing and only 5% reported concerns with self-sampling.27 Another study in Kenya demonstrated that self-collection is just as efficacious as physician-collected samples for detection of high-grade cervical precancer [cervical intraepithelial neoplasia (CIN)2+].28 A recent meta-analysis found that self-collection had a lower sensitivity than clinician samples to detect CIN2+ (77% vs 93%), but that sensitivity was 96% for both sampling methods when PCRs were used.29 Self-collection for HR-HPV could be used to triage women for confirmation with other screening methods,30 and any reduction in sensitivity with self-testing would likely be offset by increased coverage. With only an estimated 8% of Kenyan women aged 25–45 having ever been screened,9 better screening programmes need to be created to reach women who are most in need, including high-risk women who are living with HIV.
This study benefited from a relatively large sample of HIV-infected women with detailed sociodemographic, sexual behaviour and physical examination data collected for both HIV-infected women and their uninfected male partner. Despite this, interpretation of these results is limited to ART-naive, HIV-infected women in an HIV-discordant relationship, and our findings may not be generalisable to other HIV-seropositive women, including those on ART. We chose to use PRs to measure the strength of associations between factors of interest and the presence of HPV infections and abnormal cytology because these PRs can be directly interpreted as the relative prevalence of the ‘outcome’ when the outcome is common (eg, detection of HPV) or rare (eg, high-grade lesions). As unadjusted cross-sectional measures of associations, these findings cannot be interpreted as causal relationships. Despite the sample size and relatively high rate of high-grade lesions, the absolute number of high-grade abnormal cytology cases (n=16) and SCC (n=2) was small. Moreover, our outcome assessment was based on conventional cytology without histological confirmation of high-grade lesions because only a subset of women with an HSIL cytology finding returned for colposcopy in a timely manner; however, cytological results are relevant as they are used in clinical practice for screening and referral to colposcopy.
In summary, we found a high burden of HPV infection among HIV-infected women living in Kenya that was strongly associated with the development of high-grade cervical lesions. The distribution of some HR-HPV types among HIV-infected Kenya women with high-grade lesions differs from other geographical regions and elsewhere in Africa, in particular an enriched prevalence of HPV 31, 35 and 58, prompting a need for investigation of their contribution to cervical cancer and their relevance to screening in this population. Self-testing or future point-of-care testing optimised to the region-specific HPV genotypic distribution should be considered to reduce the high burden of cervical cancer in Africa.
Key messages
Nearly half of HIV-infected Kenyan women in discordant relationships were infected with a high-risk human papillomavirus (HR-HPV) genotype.
Screening for HR-HPV genotypes would identify a large majority of women who have a high-grade cervical lesion or more severe cytology.
HR-HPV screening would also yield positive results for a large number of women with normal or low-grade cervical cytology, necessitating a well-designed secondary screening process.
Acknowledgments
We would like to thank the staff at the Couples Against Transmission study centres in Nairobi, Kenya and the women and men who volunteered to participate in this study. We would also like to thank Hologic (Marlborough, Massachusetts) for donating ThinPrep liquid for this study.
References
Supplementary materials
Supplementary Data
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Footnotes
Handling editor Francesca Ceccherini Silberstein
Contributors BG, AFR, CF, RB, RC, JK and JSS conceived and planned the study. RB and RC conducted study recruitment and collected the specimens. AFR, JAC, RB, RC and JSS contributed to sample preparation and conducted the assays. BG, AFR and JAC planned and carried out the analyses. BG wrote the manuscript with support from AFR, JAC and JSS. All authors provided critical feedback, helped shape the research, analysis and manuscript, and approved the final version to be published.
Funding Research support included National Institutes of Health (NIH) (R01 AI0684316 and R24 TW007988) from Fogarty International Center (FIC) through Vanderbilt University. In addition, CF received support from NIH (K24 AI087399), RB was a scholar at the University of Washington International AIDS Research and Training Program (IARTP) funded by FIC (D43 TW000007), and BG was supported by an NIH/FIC postdoctoral fellowship (3D43TW000007-22S1) and by an NIH/NIAID Career Development Award (K01AI098527). AFR was supported by an NIH/NICHD Career Development Award (K23HD071788). RC received support from the International Research Scientist Development Award (K01 TW008406). This publication was made possible with help from the University of Washington Center for AIDS Research (CFAR), an NIH-funded programme (P30 AI027757), and the University of North Carolina at Chapel Hill Center for AIDS Research (CFAR) (P30 AI050410), which are supported by the following NIH institutes and centres: NIAID, NCI, NIMH, NIDA, NICHD, NHLBI and NIA.
Disclaimer The content of this manuscript is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health, Fogarty International Center or Vanderbilt University.
Competing interests JSS has received unrestricted grants, research support or consultancy fees from Hologic, BD Corporation and Trovagene within the last 5 years.
Patient consent for publication Not required.
Ethics approval Written informed consent was obtained and ethical approval was granted by the Institutional Review Board (IRB) at the University of Washington (#30243) and the Ethics and Research Committee (ERC) at Kenyatta National Hospital (P25/2/2007).
Provenance and peer review Not commissioned; externally peer reviewed.
Data availability statement Data are available upon reasonable request. Use of data is subject to approval of the Kenyatta National Hospital Ethics and Research Committee.