Twelve-Year Trend in the Prevalence of High-Risk Human Papillomavirus Infection Among Rwandan Women Living With HIV

Abstract

Background

We examined the trend in prevalence of high-risk human papillomavirus (hrHPV) cervical infection among Rwandan women living with HIV (WLWH) over 12 years.

Methods

Prevalence of cervical hrHPV DNA was measured in 3 studies at 3 different time periods in 3 different groups of WLWH using 3 different but comparable hrHPV tests: a MY09/MY11 PCR test in 2005 (RWISA; n = 497), careHPV in 2009–2010 (HPV Demonstration; n = 1242), and Xpert HPV test in 2016–2018 (U54; n = 4734). Prevalences were adjusted for age and CD4 cell count.

Results

HrHPV prevalence decreased over time from 42.5% to 32.2% to 26.5% (P < .001). CD4 cell counts improved over time (P_trend <.001) so that the percentage of WLWH with CD4 counts of ≥500 cells/μL increased from 7.7% in 2005 to 42.2% in 2009–2010 and 61.1% in 2016–2018. Thus, after adjustment for differences in CD4 counts and age, hrHPV prevalences were more similar over time: 32.6% for RWISA, 30.6% for HPV Demonstration, and 27.1% for U54 (P = .007).

Conclusions

Prevalence of hrHPV among WLWH has decreased over the past decade, most likely the result of improved immune reconstitution due to better HIV care and management in Rwanda.

We observed that the age-specific prevalence of high-risk human papillomavirus in Rwandan women living with human immunodeficiency virus has decreased over a 12-year period as immune reconstitution has improved due to better HIV care.

Infection with high-risk human papillomavirus (hrHPV) is recognized as one of the major causes of infection-related cancer in both men and women. HPV infections cause virtually all cervical cancer, the fourth most common cancer and fourth leading cause of cancer-related deaths among women worldwide [1]. It also causes most anogenital cancers, for example, of the vulva, vagina, penis, and anus, as well as a significant proportion of head and neck cancers [2, 3]. HPV is responsible for 4.5% (630 000) of all new cancer cases worldwide [4].

The burden of cervical cancer is especially high in sub-Saharan Africa, which has the highest prevalence of human immunodeficiency virus (HIV) and acquired immune deficiency syndrome (AIDS) of any region globally [5, 6]. Cervical cancer was included as an AIDS-defining disease in adolescents and adults in 1993 [4–8]. Women living with HIV (WLWH) have significantly elevated risk of cervical cancer [9–11]. In Rwanda, one study found WLWH to be at a 6-fold greater risk of cervical cancer than HIV-negative women [12]. WLWH have a greater risk of HPV acquisition, prevalence, and persistence, and progression to cervical precancer than HIV-uninfected women, with the risk inversely related to degree of immunosuppression as measured by the CD4 count [13].

In 2005, the Rwandan Ministry of Health initiated an aggressive program to limit the spread and ultimately reduce the burden of HIV in Rwanda [14]. By 2017, the Rwandan Ministry of Health introduced a treat all policy for people living with HIV (PLWH) in line with the 2015 WHO recommendations that all PLWH be treated with antiretroviral therapy (ART) as soon as possible [15].

In 2011–13, Rwanda, through a donation from Merck, launched a national HPV vaccination program. In 2011, over 92 000 girls in primary school grade 6 (approximately 12 years old) were vaccinated with 6 doses of Gardasil [16]. During 2012 and 2013, a catch-up vaccination targeted girls in secondary school grade 3 (approximately 15 years old). In 2014, HPV vaccination was supported by GAVI and it reverted to 12 year old girls [16, 17]. In 2015, Rwanda switched to vaccinating only 12 year olds with 2 doses at 6 months apart. In all years, Rwanda achieved ≥90% annual coverage with the recommended number of HPV vaccine doses in the target population [18–20]. Thus, Rwanda was one of the earliest and most successful adopters of HPV vaccination globally.

During a period from 2005 to 2018, we conducted 3 separate studies of HPV. In 2005, we found in a population of WLWH enrolled in the Rwanda Women’s Interassociation Study and Assessment (RWISA) a prevalence of hrHPV infection among WLWH of 46%, which was strongly inversely associated with concurrent CD4 counts [21]. In 2010, we found in an HPV testing demonstration project (HPV Demonstration) a prevalence of hrHPV infection among WLWH of 32%, which was likewise strongly inversely associated with concurrent CD4 counts [22]. In 2016–2018, we conducted a study sponsored by a U54 grant from the US National Cancer Institute (U54) of cervical screening methods in WLWH, for which the hrHPV prevalence and its relationship with CD4 counts has not yet been reported. This gave us a unique opportunity to report the hrHPV prevalence in the latter study and describe the time trends in hrHPV. For the first time, we were able to investigate the impact of population HIV control on hrHPV infection, which is the obligate viral cause of nearly all cervical cancer. Given the role of HIV in cervical cancer and that the degree of immunosuppression is related to the early natural history of HPV, we hypothesized that hrHPV prevalence has declined because of improved immune function as measured by CD4 counts and, perhaps secondarily, as the result of comprehensive HPV vaccination. We compared the cervical hrHPV prevalence, which is correlated with cervical cancer incidence [23] and therefore a marker of population risk, across the 3 studies.

METHODS

RWISA

RWISA was an observational prospective cohort study of 710 WLWH and 226 HIV-uninfected Rwandan women living in Kigali as previously described in detail [21]. Briefly, the enrollment and baseline assessment were conducted in 2005. Enrolled women completed an administered questionnaire, a physical examination during which blood was collected for determination of CD4 cell counts, and a pelvic examination during which a 10-mL cervical vaginal lavage (CVL) using buffered saline was collected for HPV genotyping.

CVLs were tested for HPV genotypes 6, 11, 13, 16, 18, 26, 31–35, 39, 40, 42, 45, 51–59, 61, 62, 64, 66–74, 81–85, 89, and 97 using the L1 MY09/MY11 modified polymerase chain reaction (PCR) system with AmpliTaq Gold polymerase as described elsewhere [21]. Briefly, the specimens were tested for HPV genotypes with the following steps: (1) aliquots of specimens were digested, (2) the HPV DNA in the cellular digest was amplified using the L1 MY09/MY11 primers with AmpliTaq Gold polymerase, (3) amplification products were probed for the presence of any HPVDNA by Southern blot analysis with a radiolabeled generic probe mixture for HPV, and (4) those positive for HPV were typed by dot blot hybridization for the aforementioned specific HPV types.

HPV Demonstration

In 2009–2010, a cross-sectional study of HPV testing for cervical screening was conducted in Rwanda using careHPV (Qiagen) [22]. A sample of 2964 eligible, consenting women, stratified on HIV status, aged 30–60 years were enrolled. Women were recruited from Nduba and Jabana sectors served by the Nyacyonga Health Center under the Kibagabaga District Hospital in Gasabo District, with 1000 of the WLWH recruited from an HIV-care–dedicated clinic in Kigali. Enrolled women were administered a short questionnaire, tested for HIV if HIV status was unknown, and if positive blood was drawn for CD4 cell counts. They then underwent a pelvic examination during which a cervical sample was collected in DCM medium (Qiagen) for careHPV testing and visual inspection after acetic acid application (VIA) was performed. Of the 2964 women enrolled, 1289 were WLWH and 1675 HIV-uninfected women, aged 30–60 years.

Cervical specimens in DCM buffer were tested by careHPV according to the manufacturer’s instructions [24]. careHPV is lower-cost signal amplification DNA test based on Hybrid Capture 2 (Qiagen) assay technology for a pool of high-risk HPV genotypes (HPV16, 18, 31, 33, 35, 39, 45, 51, 52, 56, 58, 59, 66, and 68). careHPV does not provide any readout for individual or smaller pools of HPV genotypes.

U54

In 2016–2018, a cross-sectional study was conducted to examine different strategies for cervical screening of WLWH [25]. A total of 5023 WLWH living in the Kigali area and attending routine HIV care were enrolled into the study. Enrolled women completed an administered short questionnaire and underwent a pelvic examination, at which time a cervical specimen was collected into PreservCyt medium (Hologic) for Xpert HPV testing (Cepheid), VIA performed, and a digital image of the cervix was collected using an EVA system (MobileODT). Data on HPV vaccination were not collected for this cohort of women as they were almost a decade older than the cohort of Rwandan women first vaccinated against HPV [16, 19].

Cervical specimens in PreservCyt were tested using the Xpert HPV test, a qualitative, real-time PCR assay for the detection of hrHPV DNA per the manufacturer’s instructions [26]. The Xpert HPV assay includes simultaneous detection of 14 hrHPV types, hydroxymethylbilane synthase, and an internal probe check control. The 14 targeted hrHPV genotypes (HPV16, 18, 31, 33, 35, 39, 45, 51, 52, 56, 58, 59, 66, and 68) are detected in 5 fluorescent channels: (1) HPV16; (2) HPV18 and 45; (3) HPV31, 33, 35, 52, and 58; (4) HPV51 and HPV59; and (5) HPV39, 56, 66, and 68.

Analysis

The primary outcome of the analysis was the prevalence of hrHPV, defined as positive for hrHPV genotypes 16, 18, 31, 33, 35, 39, 45, 51, 52, 56, 58, 59, 66, and/or 68, which all 3 HPV tests detected individually (MY09/11), as a pool (careHPV), or individually or as small pools of hrHPV genotypes (Xpert). Age was grouped as 30–34, 35–39, 40–44, 45–49, and 50–54 years; women younger or older than 30–54 years were excluded from this analysis because there were few women enrolled with these ages and not in every study. Blood CD4 counts were categorized as <200, 200–349, 350–499, and ≥500 cells/μL. Analyses were restricted to WLWH who had available hrHPV test results, age, and CD4 counts data.

hrHPV prevalence was calculated for each study, overall and stratified by age group and CD4 categories. Differences across studies were tested for statistical significance using Fisher exact test. Differences across age groups and CD4 categories were tested for statistical significance using a nonparametric test of trend [27].

Unadjusted hrHPV prevalence and binomial 95% confidence intervals (CI) for each study were calculated and differences across studies were tested for statistical significance using Fisher exact test. Logistic regression models were used to calculate age group-adjusted, CD4 category-adjusted, and age group-adjusted and CD4 category-adjusted hrHPV prevalence and 95% CI for each study. Odds ratio (OR) and 95% CI for hrHPV prevalence, with RWISA as the reference group, as well as testing for statistically significant differences in the adjusted hrHPV prevalence were calculated using the same logistic regression models. For RWISA and U54, the unadjusted prevalence and age group-adjusted and CD4 category-adjusted prevalence of the hrHPV groups defined by Xpert HPV testing channels, as well as OR, 95% CI, and testing for statistically significant differences, were likewise calculated.

Analyses were done using STATA (version 15.1; STATA Corporation) and P values less than .05 were considered statistically significant.

All subjects provided written, informed consent. The Rwanda National Ethics Committee and the Montefiore Medical Center/Albert Einstein College of Medicine Institutional Review Board approved all 3 studies.

RESULTS

A summary of the 3 studies is found in Table 1. A total of 497 of 647 (76.8%) WLWH from RWISA, 1242 of 1289 (96.4%) WLWH from HPV Demonstration, and 4734 of 5007 (94.5%) WLWH from U54 were included in these analyses. The RWISA cohort was the youngest and had the lowest CD4 counts and the U54 cohort was the oldest and had the highest CD4 counts (P_trend < .001 for both). Notably, the percentage of women aged 50–54 years was 2.2% in RWISA, 7.4% in the HPV Demonstration, and 10.6% in the U54 cohorts. The percentage of women with CD4 counts of ≥500 cells/μL was 7.7% in RWISA, 42.2% in the HPV Demonstration, and 61.1% in the U54 cohorts. All WLWH in RWISA were naive to ART while 22.3% of WLWH in HPV Demonstration and 100% of WLWH in U54 were on ART at the time of their enrollment.

Table 1.

Description of the 3 Studies of hrHPV in WLWH in Kigali, Rwanda

Characteristic	RWISA [21]	HPV Demonstration [22]	U54 [25]
Study year	2005	2009–2010	2016–2018
HPV test	MY09/MY11 PCR	careHPV	Xpert
No. enrolleda	647	1289	5007
No. with resultsb	497	1242	4734
Age, y, mean, median (IQR)	37, 36 (33–40)	40, 39 (35–44)	41, 40 (35–45)
CD4 cells/μL, mean, median (IQR)	271, 250 (162–765)	448, 477 (322–607)	679, 570 (414–747)
Age group, y, No. (%)c
30–34	213 (42.9)	292 (23.5)	996 (21.0)
35–39	144 (29.0)	374 (30.1)	1253 (26.5)
40–44	82 (16.5)	306 (24.6)	1102 (23.3)
45–49	47 (9.5)	178 (14.3)	864 (18.3)
50–54	11 (2.2)	92 (7.4)	519 (11.0)
CD4 cells/μL, No. (%)c
<200	175 (35.2)	87 (7.0)	264 (5.6)
200–349	200 (40.2)	278 (22.4)	549 (11.6)
350–499	84 (16.9)	353 (28.4)	1029 (21.7)
≥500	38 (7.7)	524 (42.2)	2862 (61.1)
Receiving ART, No. (%)c
No	497 (100)	948 (76.3)	0 (0)
Yes	0 (0)	277 (22.3)	4733 (100)
Missing data	0 (0)	17 (1.4)	1 (0)

Abbreviations: ART, antiretroviral therapy; hrHPV, high-risk human papillomavirus; IQR, interquartile range; PCR, polymerase chain reaction; RWISA, Rwanda Women’s Interassociation Study and Assessment; WLWH, women living with human immunodeficiency virus.

^aTotal number of WLWH enrolled.

^bNumber of WLWH aged 30–54 years with CD4 count data and HPV test results; RWISA, 148 were missing data on age or were outside the 30–54-year age range and 2 were missing data on CD4; HPV Demonstration, 45 were missing data on age or were outside the 30–54-year age range and 11 were missing data on CD4 counts; U54, 10 were missing HPV data and among those with HPV data, 41 were missing data on age or were outside the 30–54-year age range, 219 were missing data on CD4 counts, and 3 were missing data on age or were outside the 30–54-year age range and CD4 counts.

^c P < .001.

HrHPV prevalence decreased in older age groups (P_trend = .074 for RWISA; P_trend < .001 for HPV Demonstration and U54) and with higher CD4 count categories (P_trend = .001 for RWISA; P_trend = .049 for HPV Demonstration; P_trend < .001 for U54; Table 2). There were significant differences in hrHPV prevalence within age groups of 30–34 (P < .001), 35–39 (P < .001), and 40–44 (P = .007) years but not in the other age groups. These differences between cohorts in hrHPV prevalence within age groups may be partially explained by the differences in the CD4 counts across cohorts within each age group (P < .001 for all).

Table 2.

hrHPV Prevalence in WLWH by Study, Overall and Stratified by Age Group and CD4 Counts

	HPV Positive, No. (%)			P a
	RWISA	HPV Demonstration	U54
Overall	211 (42.5)	399 (32.1)	1248 (26.4)	< .001
Age group, y
30–34	96 (45.1)	136 (46.6)	300 (30.1)	< .001
35–39	66 (45.8)	104 (27.8)	362 (28.9)	< .001
40–44	30 (36.6)	89 (29.1)	258 (23.4)	.007
45–49	14 (29.8)	47 (26.4)	200 (23.2)	.386
50–54	5 (45.5)	23 (25.0)	128 (24.7)	.303
P trend b	.07	< .001	< .001
CD4 cells/μL
<200	93 (53.1)	40 (46.0)	119 (45.1)	.240
200–349	75 (37.5)	89 (32.0)	188 (34.2)	.460
350–499	34 (40.5)	109 (30.9)	263 (25.6)	.005
≥500	9 (23.7)	161 (30.7)	678 (23.4)	.002
P trend b	.001	.049	< .001

Abbreviations: HPV, human papillomavirus; RWISA, Rwanda Women’s Interassociation Study and Assessment; WLWH, women living with human immunodeficiency virus.

^aFisher exact test.

^bTest of trend [21].

Table 3 shows the unadjusted hrHPV, age group-adjusted, CD4 category-adjusted, and age group and CD4 category-adjusted hrHPV prevalences. Unadjusted hrHPV prevalences were 42.5% (95% CI, 38.2%–46.8%) for RWISA, 32.1% (95% CI, 29.6%–34.8%) for HPV Demonstration, and 26.4% (95% CI, 25.1%–27.6%) for U54 populations (P < .001). Age group-adjusted hrHPV prevalences were 40.4% (95% CI, 36.1%–44.8%) for RWISA, 31.7% (95% CI, 29.2%–34.4%) for HPV Demonstration, and 26.5% (95% CI, 25.2%–27.8%) for the U54 populations (P < .001). CD4 category-adjusted hrHPV prevalences were 34.8% (95% CI, 30.5%–39.3%) for RWISA, 31.0% (95% CI, 28.5%–33.7%) for HPV Demonstration, and 27.0% (95% CI, 25.7%–28.3%) for the U54 populations.

Table 3.

Overall hrHPV Prevalence, Unadjusted and Adjusted for Age Group, CD4 Count Category, and Both

	RWISA		HPV Demonstration				U54				P
	% HPV+	95% CI	% HPV+	95% CI	ORa	95% CI	% HPV+	95%CI	ORa	95% CI
HPV unadjusted	42.5	38.2–46.8	32.1	29.6–34.8	0.64	.52–.79	26.4	25.1–27.6	0.49	.40–.59	< .001
Age group-adjusted HPV	40.4	36.1–44.8	31.7	29.2–34.4	0.69	.55–.85	26.5	25.2–27.8	0.53	.44–.64	<.001
CD4 category-adjusted HPV	34.8	30.5–39.3	31.0	28.5–33.7	0.84	.68–1.06	27.0	25.7–28.3	0.69	.56–.85	< .001
Age group and CD4 category-adjusted HPV	32.6	28.4–37.1	30.6	28.1–33.3	0.91	.73–1.14	27.1	25.8–28.4	0.77	.62–.95	.007

Abbreviations: CI, confidence interval;HPV, human papillomavirus; OR, odds ratio;RWISA, Rwanda Women’s Interassociation Study and Assessment.

^aVersus RWISA.

Adjustment for age group and CD4 category were both significant (P < .001) and the differences in age group- and CD4-adjusted hrHPV prevalence across studies, while still significant (P = .007), were relatively minor: 32.6% (95% CI, 28.4%–37.1%) for RWISA, 30.6% (95% CI, 28.1%–33.3%) for the HPV Demonstration, and 27.1% (95% CI, 25.8%–28.4%) for the U54 populations. The likelihood of being hrHPV positive compared to RWISA and adjusted for age group and CD4 category was OR 0.91 (95% CI, .73–1.14) for HPV demonstration and OR 0.77 (95% CI, .62–.95) for U54.

Unadjusted and age group- and CD4 category-adjusted prevalences of HPV groups, defined by the Xpert HPV test channels, were compared between RWISA and the U54 (Table 4). The difference in the unadjusted prevalences of HPV16, the most carcinogenic HPV genotype, was significant (9.7% vs 5.9%, P = .002) but the difference in the adjusted HPV16 prevalences was much smaller and not statistically significant (7.4% vs 5.9%, P = .209). The difference in the unadjusted prevalences of HPV18 and HPV45 combined, the second and third most carcinogenic HPV genotype, was significant (8.5% vs 5.4%, P = .008) but the difference in the adjusted HPV18 and HPV45 prevalences was negligible (6.0% vs 5.5%, P = .633).

Table 4.

Comparison of Age Group and CD4 Category Adjusted Prevalence of HPV Groups in RWISA and U54 studies

HPV Group	RWISA		U54		Unadjusteda		Adjusted b
	Unadjusted, % (95% CI)	Adjusted, % (95% CI)	Unadjusted, % (95% CI)	Adjusted, % (95% CI)	OR (95% CI)	P c	OR (95% CI)	P
16	9.7 (7.2–12.6)	7.4 (5.4–10.2)	5.9 (5.2–6.6)	5.9 (5.3–6.6)	0.58 (.42–.80)	.002	0.79 (.55–1.14)	.209
18/45	8.5 (6.2–11.3)	6.0 (4.2–8.4)	5.4 (4.8–6.1)	5.5 (4.8–6.1)	0.62 (.44–.87)	.008	0.91 (.62–1.34)	.633
31/33/35/52/58	23.5 (19.9–27.5)	15.7 (12.7–19.2)	14.0 (13.0–15.0)	14.1 (13.1–15.1)	0.53 (.42–.66)	<.001	0.88 (.69–1.14)	.335
51/59	9.3 (6.9–12.2)	5.9 (4.2–8.4)	3.6 (3.1–4.2)	3.6 (3.1–4.2)	0.37 (.26–.52)	<.001	0.60 (.40–.89)	.012
39/56/66/68	10.5 (7.9–13.5)	5.4 (3.8–7.5)	6.1 (5.4–6.8)	6.0 (5.3–6.7)	0.56 (.41–.76)	<.001	1.10 (.78–1.56)	.586

The HPV groups were defined based on the read-out channels for the Xpert HPV test that was used in the U54 study. HPV genotype-specific data from a validated PCR assay used in the RWISA study were combined to mimic these HPV groupings.

Abbreviations: CI, confidence interval; HPV, human papillomavirus; OR, odds ratio; RWISA, Rwanda Women’s Interassociation Study and Assessment; WLWH, women living with human immunodeficiency virus.

^aVersus RWISA.

^bVersus RWISA (adjusted for age group and CD4 category).

^cFisher exact test.

DISCUSSION

We observed a significant, approximately 50% decline in the age-specific hrHPV prevalence in 3 separate cohorts of WLWH women over an approximately 12-year period. To our knowledge, this is the first report of such an effect on hrHPV. Differences in the age of the cohorts only partially explained the differences in hrHPV prevalence. The main explanatory difference between the cohorts was the difference in immune reconstitution, as measured by CD4 counts, between the 3 cohorts. From our convenience samples of WLWH over time, their immune competence has improved, as evident from the higher CD4 counts and, as a consequence, lowered the prevalence of hrHPV. Once hrHPV prevalences were adjusted for age and CD4, the magnitude of these differences was greatly diminished.

Our results are consistent with a recent meta-analysis reporting that ART use was associated with lower hrHPV prevalence [13]; ART use was associated with hrHPV prevalence in this study (P < .001, data not shown) but not after adjustment by CD4 categories. An analysis from a cohort consortium of North American sites found an inverse relationship of CD4 counts and the risk of cervical cancer [11], suggesting that that HIV control will have a significant population impact on cervical cancer incidence. Population HPV prevalence is correlated with cervical cancer incidence [23].

In 2018, the World Health Organization (WHO) called for the elimination of cervical cancer as a public health problem [28]. The WHO strategic plan includes achieving 70% coverage for cervical screening of midadult women globally [29]. One important implication of these and other data [13] is that WLWH under HIV care will be less likely to test positive for hrHPV or any other screening test in cervical screening programs, which is especially important in sub-Saharan Africa where there is limited clinical infrastructure to manage and treat those screen-positive WLWH at risk for cervical cancer.

Notably, there remained a small, unexplained, residual difference in hrHPV prevalence over time. We offer several possible explanations. First, it is possible that the HPV vaccination program with Gardasil (Merck & Co.), started in 2011 by vaccinating grade 6 girls (12 years old) and achieving > 90% coverage [19], has reduced the population carriage of HPV16 and HPV18, the 2 targeted hrHPV genotypes. This would not be a direct effect of HPV vaccination on the U54 cohort, as those who received HPV vaccination still would only be at their oldest about 20 years of age in 2018 and thus not included in the U54 cohort, but rather due to herd protection as has been seen in Australia [30]. While the difference in adjusted prevalence of HPV16 was 1.5%, there was no significant difference in HPV18/45, and there were major differences in adjusted prevalence for HPV genotype groups HPV31/33/35/52/58 and HPV51/59, genotypes not targeted by Gardasil. Thus, the hrHPV trends over time, and residual differences after adjustment for age and CD4 counts, seem unlikely to be explained fully by HPV vaccination by Gardasil.

A second possibility is that improved population general immunity in PLWH due to immune reconstitution has reduced the endemicity of hrHPV in these populations, thereby reducing incident hrHPV infection. We cannot rule out this explanation but this would assume a closed sexual network of PLWH among some of the women in our studies. Unfortunately, we did not survey the male partners of the WLWH in these 3 cohorts, which would allow us to explore this possibility.

Perhaps the most likely explanation is that CD4 counts is a good but imperfect measure of immune status and the residual difference in hrHPV prevalence between studies is due to unmeasured differences in immune competence. We did not have concurrent HIV viral load for all 3 studies, nor did we have the nadir CD4. Indeed, HIV viral load is now the recommended method to monitor the control of HIV infection by ART [31]. HIV viral load data were only available in the U54 study; HIV viral load (log units) was independent associated with hrHPV after adjustment for CD4 category and age group (data not shown). Nadir CD4 reportedly is associated with the risk of cervical precancer and cancer [32], presumably affecting the likelihood of an hrHPV infection persisting, which is a prerequisite, and strong predictor of, developing cervical precancer and cancer [2, 33, 34]. There are also other immune markers, specifically regulatory T cells (Tregs) that have been shown to be associated with HPV and its natural history [35, 36].

It is also worth noting that different assays were used in all 3 studies, which may have introduced some differences in the measurements of hrHPV prevalence. Two validated PCR-based assays were used in the earliest and latest measurements and we expect these to be the more analytically sensitive tests. careHPV testing might be expected to have the least analytic sensitivity for hrHPV and therefore, had a PCR-based assay been used, the hrHPV prevalence measured in HPV Demonstration might have been closer to the hrHPV prevalence in RWISA. However, careHPV benchmarks closely with Hybrid Capture 2 [37, 38], a long-standing industry standard for HPV testing with which MY09/11 [39] and Xpert [40] assay have good agreement. careHPV also has shown good agreement with the MY09/11 [41]. Thus, while the different assays used could have introduced small differences in hrHPV prevalence across studies, it is unlikely to explain time trends in hrHPV prevalence or even fully explain the residual differences in adjusted hrHPV prevalences.

In addition to the limitation of using different assays in the 3 studies, another limitation for these studies was the use of convenience samples of the WLHW population. True random sampling of the population was not feasible. However, given that hrHPV infection is asymptomatic unless cancer is present, it seems unlikely that convenience sampling of the WLWH population would have introduced a systematic bias that could explain our findings. Notably, our data from the U54 cohort are comparable to the hrHPV prevalence previously reported for Rwandan WLWH aged 35 years and older [16].

However, it is possible that those WLWH who are sick with AIDS may not have attended clinics from which we would have recruited them to participate. If that is the case, then the true population hrHPV prevalence likely would be higher, but higher in all 3 studies. Indeed, if the population health of the WLWH has improved over time, as we have surmised, then there would have been more WLWH with AIDS in the earlier studies who might not have attended their care and therefore the unadjusted differences in hrHPV prevalence might be greater still. However, in 2005, WLWH with AIDS may have been more likely to participate as they were desperate for ART and the site of care and enrollment for RWISA was one of the few places where ART was available. In this case, the true population hrHPV prevalence in WLWH could have been lower than what we report.

In conclusion, the hrHPV prevalence in Rwandan WLWH has decreased as their health and immune status have improved secondary to improving HIV care, treatment, and management as well as early initiation of ART in Rwanda. Under the treat all era, a recommendation by the WHO to treat all PLWH as soon as they are diagnosed regardless of their CD4 cell count and clinical stage, nearly all WLWH (approximately 95%) are on ART [42]. The reduction in hrHPV carriage in this population could reduce the burden of cervical cancer in WLWH living in Rwanda due to reduced hrHPV incidence and hrHPV persistence, the latter critical for the development of cervical precancer and cancer [34]. However, improved immunity may have an unintended consequence of increased risk of cervical cancer due to reduced competing HIV-related mortality and increased life expectancy, allowing sufficient time to develop HPV-related cancers [10, 43–46]. Given the call for the elimination of cervical cancer as a public health problem by the WHO, our data, along with recent meta-analyses [13], highlight the importance of HIV treatment and immune reconstitution for the prevention of cervical cancer in WLWH. Programs focused on cervical cancer elimination in regions of high HIV burden should emphasize the diagnosis and treatment of HIV in women for their overall health as well as reducing their risk of cervical cancer and cervical cancer-related mortality. However, concomitant improvement in access to screening for the prevention of cervical cancer in WLWH living in sub-Saharan Africa may be critical to realize those benefits of better natural immunity to hrHPV and will be necessary to achieve the aforementioned WHO goals.

Notes

Acknowledgments. We acknowledge the great contribution of the study participants as well as all the study team, including nurses, laboratory technicians, data manager, logistics team, and the administration and finance team.

Financial support. This work was supported by the National Institutes of Health (NIH) National Cancer Institute (grant number 5U54CA19016304) and NIH National Institute of Allergy and Infectious Diseases (grant number 5U01AI096299-09).

Potential conflicts of interest. P. E. C. has received cervical screening tests and diagnostics at a discounted or no cost for research from Roche, BD, Cepheid, and Arbor Vita Corporation; Qiagen donated careHPV tests; and Cepheid provided Xpert tests at a discounted cost and a loaner GX16 system for these studies. All other authors report no potential conflicts.

All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.

Presented in part: 11th African Organization for Research and Training in Cancer, Kigali, Rwanda, 7–11 November 2017; 33rd International Papillomavirus Conference, Sydney, Australia, 2–6 October 2018; and several regional conferences.