Cervical human papillomavirus DNA detection in women living with HIV and HIV-uninfected women living in Limbe, Cameroon

Abstract

Background.

There are limited data on cervical HPV prevalence in Cameroon and none from its Anglophone region. We investigated cervical HPV prevalence in HIV-uninfected (HIV[−]) and HIV-infected (WLWH) women living in the region.

Methods.

A convenience sample of consecutively recruited HIV[−] women (n=295) and women living with HIV (WLWH) (n=560) attending the Limbé Regional Hospital were enrolled into a cervical screening study. Women underwent screening that included HPV testing of self-collected and provider-collected specimens. We calculated the HPV prevalence by HIV status, overall and stratified by age, and among WLWH, stratified by CD4 counts. We compared the concordance for the detection of HPV between self- and provider-collected specimens.

Results.

Crude HPV prevalence was 21.69% (95% confidence interval [95%CI]=17.21–26.48%) for HIV[−] women and 46.43% (95%CI=42.24–50.66%) for WLWH (p<0.001). Among WLWH, older age (p_trend=0.01) and higher CD4 counts (p_trend=0.007) were associated with lower HPV prevalence. There was a good-to-excellent agreement for HPV detection between specimens, and self-collected were more likely than provider-collected specimens to test HPV positive, for all women and stratified by HIV status.

Conclusions.

HIV-related immunosuppression was a risk factor for HPV prevalence in this population. HPV testing of self-collected specimens appeared to be less specific than HPV testing of provider-collected specimens.

Introduction

Women in Cameroon have limited access to cervical cancer prevention and treatment programs due to weak health infrastructure and costs associated with Pap smear among other barriers. Recently, World Health Organization¹ recommended human papillomavirus (HPV) and/or visual inspection after acetic acid (VIA) as the preferred screening methods in places where high-quality cervical cytology/Pap testing is not available.

To date, there are limited data on different cervical screening strategies in Cameroon, a high cervical-cancer burden country in Central sub-Saharan Africa (SSA). Cameroon has age-standardized cervical cancer incidence rate of 31.3 per 100,000 (28^th in the world) and mortality rate of 21.9 per 100,000 (26^th in the world).² The high rates of cervical cancer incidence and mortality are due in part to the high prevalence of HIV, which is estimated at 4.7% [95% CI: 4.1%−5.3] in 2018.³

One group has conducted and reported on a series of studies on HPV detection in the central, Francophone regions of Cameroon.^4–7 While those studies provide an important reference for the feasibility of implementing HPV testing-based screening there, there are limited data on testing of women living with HIV (WLWH) in Cameroon and no data for WLWH and HIV-uninfected (HIV[−]) women in the Anglophone regions that borders Nigeria. To address this gap in knowledge, we conducted a pilot study of cervical screening strategies in the coastal town of Limbé, Cameroon. Here, we report on the HPV prevalence and the concordance of HPV testing on self-collected cervicovaginal and provider-collected cervical specimens.

Methods

Population.

We recruited a convenience, consecutive sample of women attending the Limbé Regional Hospital in Limbé, Cameroon; WLWH were recruited at the HIV Treatment Center and HIV[−] women were recruited from the Outpatient Clinic. Our goal was to recruit a stratified sample of two WLWH to every one HIV[−] woman. Inclusion criteria were aged 25–59 years, confirmed to be either a WLWH or HIV[−] woman, never undergone cervical cancer screening, no history of invasive cervical cancer, and willing, and able to competently understand and provide written, informed paper-based consent to participate. Exclusion criteria were those who did not meet the inclusion criteria or were pregnant, had signs of abnormalities or non-menstrual bleeding suggestive of invasive cervical cancer, undergone hysterectomy, and/or, based on the judgment of the clinicians, were not sufficiently healthy to participate in a research study. Ineligible women were advised to seek routine cervical cancer screening through government programs. Women who had a menstrual period were deferred for two weeks from participating in the study. Institutional Review Board of Albert Einstein College of Medicine and the National Ethics Committee of Cameroon approved the study.

Enrollment.

All participating women provided informed, written consent to participate in the study. Consenting women completed a short nurse-administered questionnaire on risk factors and socio-demographics. Data on HIV care were extracted from medical records of those who were HIV [+]. We used Determine Test Strips⁸ to test for HIV antibodies in 220 women with previously unknown HIV status and found a 4.5% prevalence rate. Confirmatory HIV tests done at the HIV treatment Clinic also returned similar prevalence rate.

Enrolled women were escorted to a private room and provided with a device (Viba Brush),⁹ for self-collection and instructions on how to self-collect their own sample, which was then placed into ThinPrep® medium (PreservCyt; Hologic, Inc., Bedford, MA, USA). Women then underwent a pelvic exam (by a physician), at which time they had a specimen collected using Cervex Combi Brush (Rovers) into PreservCyt for ThinPrep LBC and molecular testing. Finally, VIA was be performed by a physician.

The Xpert HPV Test (Cepheid, Sunnyvale, CA, USA) on the GeneXpert platform¹⁰ was used to test the self-collected cervicovaginal specimen or the provider-collected cervical specimen for high-risk HPV per the manufacturer’s instructions.

Analyses.

We calculated the prevalence of HPV and HPV16, the only genotype detected individually by the Xpert HPV test and the most carcinogenic HPV genotype¹¹ and binomial 95% confidence intervals (95%CI) separately for HIV[−] and WLWH, overall and stratified by age group (25–29, 30–39, 40–49, and 50–59 years) and, among WLWH, by CD4 counts (per mm³) (<200, 200–349, 350–499, and ≥500). Differences in prevalence between groups of HIV[−] and WLWH were tested for statistical significance using Fisher’s exact test, and trends across strata of age and CD4 were tested for statistical significance using a non-parametric test of trend.¹² To control for age differences between the HIV[−] and WLWH, age-adjusted prevalences were calculated. Logistic regression was used to test age and CD4 groups as independent clinical determinants of HPV prevalence, given that both are well-established determinants of HPV prevalence^{11, 13–15}, and to predict the HPV prevalence for combinations of age and CD4 count.

Agreement statistics (percent agreement and kappa values) were used to compare the detection of HPV, each HPV group detected by a separate Xpert channel (HPV16, HPV18 or HPV45 (HPV16/18), HPV31, 33, 35, 52, and 58 (HPV31/33/ 35/52/58), HPV51 and HPV59 (HPV51/59), and HPV39, 56, 66, and 68 (HPV39/56/ 66/68), and HPV classified hierarchically according to cancer risk (HPV16 > HPV18/45 > other HPV >HPV negative). Differences were tested for statistical significance using an exact version of McNemar’s chi-square test.

Results

We enrolled 878 women, of which 13 (1.5%) were of unknown HIV status and excluded from further analyses. An additional 9 (1.0%) women did not have HPV testing results on the provider-collected specimen and were also excluded from these analyses. Of the 855 (97.4% of 878) women with HPV testing results that were included in these analyses, there were 295 HIV [−] women and 560 WLWH. These results included 35 specimens that were retested because of 24 error messages (including 12 retests due to lost module communication, 5 retests due to ultrasonic horn current was out of normal range, and 3 probe check failures), 10 invalid results, and 1 no result.

HIV[−] women were significantly younger than the WLWH (median of 40 vs. 42 years, respectively, p=0.006). WLWH had a mean, median, interquartile range, and range of CD4 counts of 566, 535, 329–794, and 9–2406 per mm³, respectively. Five hundred thirty-seven of 560 (95.9%) WLWH were on anti-viral therapy for their HIV at enrollment.

Table 1 shows the prevalence of HPV and HPV16 in provider-collected cervical specimens by age group and HIV status, and among WLWH, by CD4 category. Crude HPV DNA prevalences were 21.7% (95%CI=17.2–26.5%) for HIV[−] women and 46.4% (95%CI=42.2–50.7%) for WLWH (p<0.001). Crude HPV16 prevalences were 2.7% (95%CI=1.2–5.3%) for HIV[−] women and 8.8% (95%CI=6.54–11.40) for WLWH (p<0.001). Among WLWH, the prevalence of high-risk HPV (p_trend=0.007) and HPV16 (p_trend=0.1) increased with categories of decreasing CD4 counts.

Table 1.

Prevalence of high-risk human papillomavirus (HPV) and HPV16 in provider-collected cervical specimens from HIV-uninfected (HIV[−]) and HIV-infected (WLWH) women, aged 25–59 years, living in Limbé, Cameroon, overall, stratified by age group, and, for HIV+, by CD4 category. Abbreviations: N+, number of positives; %+, percent positives

		HIV[-]					WLWH					p* (HPV)	p* (HPV16)
			Any HPV		HPV16			Any HPV		HPV16
		N	N+	%+	N+	% +	N	N+	% +	N+	% +
All		295	64	21.7%	8	2.7%	560	260	46.4%	49	8.8%	<0.001	<0.001
Age Group (Years)
	25–29	39	5	12.9%	1	2.6%	24	15	62.5%	4	16.7%	<0.001	0.07
	30–39	100	27	27.0%	3	3.0%	171	89	53.2%	20	11.7%	<0.001	0.01
	40–49	97	24	24.7%	4	4.1%	256	106	42.2%	16	6.3%	0.003	0.6
	50–59	59	8	13.6%	0	0.0%	109	46	42.2%	9	8.3%	<0.001	0.03
			ptrend=0.6		ptrend=0.5			ptrend=0.01		ptrend=0.08
CD4 Counts (per mm3)
	<200						67	40	59.7%	9	13.4%
	200–349						85	43	50.6%	9	10.6%
	350–499						102	48	47.1%	8	7.8%
	≥500						306	129	42.2%	23	7.5%
								ptrend=0.007		ptrend=0.1

Prevalences of HPV (p_trend=0.01) and HPV16 (p_trend=0.08) decreased with increasing age in WLWH but there was no age trend for the prevalences of HPV (p_trend=0.6) and HPV16 (p_trend=0.5) in HIV[−] women (Table 1). Age-adjusted prevalences of high-risk HPV were 21.1% (95%CI=16.8–26.1%) in HIV[−] women compared to 46.8% (95%CI=42.6–50.9%) in WLWH (p<0.001). Age-adjusted prevalences of HPV16 was 2.5% (95%CI=1.23–5.0%) in HIV[−] women compared to 8.7% (95%CI=6.6–11.4%) in WLWH (p<0.001). In a logistic regression model, lower CD4 counts groups (p_trend=0.008) and younger age groups (p_trend=0.013) were independently associated with higher HPV prevalence among WLWH. The predicted HPV prevalence from this model ranged twofold, from 34.4% for those CD4≥500 and aged 50–59 years to 68.9% for those with CD4<200 and aged 25–29 years.

Of the 855 women enrolled in the study, 848 (99.2%) (294 HIV[−] and 554 WLWH) also had available HPV testing results on their self-collected cervicovaginal specimen. The overall agreement was 0.88 (95%CI=0.86–0.90) and the kappa was 0.76 (95%CI=0.71–0.80) for the detection of high-risk HPV in the paired provider-collected cervical and self-collected cervicovaginal specimens (Table 2). Self-collected cervicovaginal specimens were more likely to test high-risk HPV positive than the provider-collected cervical specimen (p<0.001). There was a high overall agreement (92–97%) and good-to-excellent kappa values (0.74–0.81) for the individual channels for the detection of HPV16 and pools of HPV types. The self-collected cervicovaginal specimen was more likely to test positive for each individual HPV channel although this difference was not significant for HPV16 (p=0.08) or HPV39/56/66/68 (p=0.1). The agreement statistics between provider-collected cervical specimens and self-collected cervicovaginal specimens for detection of HPV were similar in HIV[−] women (percent agreement=89%, kappa value=0.72) and WLWH (percent agreement=88%, kappa value=0.76). Self-collected cervicovaginal specimens were more likely than provider-collected cervical specimens to test positive for HPV in the HIV[−] women (26.78% vs. 21.77%, respectively, p=0.007) and WLWH (50.5% vs. 46.6%, respectively, p=0.008) (Figure). Similar patterns were observed for individual HPV channels.

Table 2.

Agreement for the detection of any human papillomavirus (HPV) and individual HPV channels for Xpert HPV testing, overall and stratified HIV status (HIV-uninfected (HIV[−]) and HIV-infected (WLWH) women), between paired provider-collected cervical and self-collected cervicovaginal specimens. Abbreviations: N+, number of positives; %+, percent positives; 95%CI, confidence interval

		Provider-Collected Specimen		Self-Collected Specimen		Positive		Positive		Negative		Negative		Percent Agreement (95%CI)	Kappa (95%CI)	P**
						Positive*		Negative*		Positive*		Negative*
		N+	%+	N+	%+	N	%	N	%	N	%	N	%
All Women (n=848)
	Any HPV	322	38.0%	359	42.3%	291	34.3%	30	3.7%	68	8.0%	458	54.0%	0.88 (0.86–0.90)	0.76 (0.71–0.80)	<0.001
	HPV16	57	6.7%	67	7.9%	49	5.8%	8	0.9%	18	2.1%	776	91.2%	0.97 (0.96–0.98)	0.77 (0.69–0.86)	0.08
	HPV18/45	70	8.3%	90	10.6%	62	7.3%	8	0.9%	28	3.3%	750	88.4%	0.96 (0.94–0.97)	0.75 (0.68–0.83)	0.001
	HPV31/33/ 35/52/58	178	21.0%	203	23.9%	158	18.6%	20	2.4%	45	5.3%	625	73.7%	0.92 (0.90–0.94)	0.78 (0.73–0.83)	0.003
	HPV51/59	60	7.1%	68	8.0%	53	6.3%	7	0.8%	15	1.8%	776	91.2%	0.97 (0.96–0.98)	0.81 (0.74–0.89)	0.1
	HPV39/56/ 66/68	101	11.9%	129	15.2%	89	10.5%	12	1.4%	40	4.7%	707	83.4%	0.94 (0.92–0.95)	0.74 (0.67–0.81)	<0.001
HIV[-] Women (n=294)
	Any HPV	64	21.8%	79	26.8%	56	19.0%	8	2.7%	23	7.8%	207	70.4%	0.89 (0.85–0.93)	0.72 (0.62–0.81)	0.007
	HPV16	8	2.7%	8	2.7%	6	2.0%	2	0.7%	2	0.7%	284	96.6%	0.99 (0.97–1.00)	0.74 (0.50–0.99)	1
	HPV18/45	14	4.8%	16	5.4%	10	3.4%	4	1.4%	6	2.0%	274	93.2%	0.97 (0.94–0.98)	0.65 (0.45–0.85)	0.5
	HPV31/33/ 35/52/58	33	11.2%	41	13.9%	29	9.9%	4	1.4%	12	4.1%	249	84.7%	0.95 (0.91–0.97)	0.75 (0.64–0.87)	0.05
	HPV51/59	9	3.1%	9	3.1%	6	2.0%	3	1.0%	3	1.0%	282	95.9%	0.98 (0.96–0.99)	0.66 (0.40–0.91)	1
	HPV39/56/ 66/68	17	5.8%	25	8.5%	16	5.4%	1	0.3%	9	3.1%	268	91.2%	0.97 (0.94–0.98)	0.74 (0.59–0.90)	0.01
WLWH (n=554)
	Any HPV	258	46.6%	280	50.5%	235	42.4%	23	4.2%	45	8.1%	251	45.3%	0.88 (0.85–0.90)	0.76 (0.70–0.81)	0.008
	HPV16	49	8.8%	59	10.6%	43	7.8%	6	1.1%	16	2.9%	489	88.3%	0.96 (0.94–0.97)	0.78 (0.68–0.87)	0.05
	HPV18/45	56	10.1%	74	13.4%	52	9.4%	4	0.7%	22	4.0%	476	85.9%	0.95 (0.93–0.97)	0.77 (0.69–0.86)	<0.001
	HPV31/33/ 35/52/58	145	26.2%	162	29.2%	129	23.3%	16	2.9%	33	6.0%	376	67.9%	0.91 (0.88–0.93)	0.78 (0.72– 0.84)	0.02
	HPV51/59	51	9.2%	59	10.6%	47	8.5%	4	0.7%	12	2.2%	491	88.6%	0.97 (0.95–0.98)	0.84 (0.76–0.92)	0.08
	HPV39/56/ 66/68	84	15.2%	104	18.8%	73	13.2%	11	2.0%	31	5.6%	439	79.2%	0.92 (0.90–0.94)	0.73 (0.66– 0.81)	0.002

^*Self-collected specimen result

^**Exact version of the McNemar chi-square test

Figure 1.

Prevalence (bars) and 95% confidence intervals (error bars) of any (high-risk) HPV type and subgroups of HPV types measured by the Xpert HPV test on paired provider-collected cervical specimen and self-collected cervicovaginal specimen from HIV-negative women and women living with HIV living in Limbé, Cameroon.

Categorizing the HPV results hierarchically according to their cancer risk¹¹, the overall agreement was 86% (95%CI=83–88%) and the kappa value was 0.75 (95%CI=0.72–0.79), with results from HPV testing of self-collected specimens being more likely to categorize women in a higher HPV-risk category (p<0.001) (Supplemental Table). Among HIV[−] (data not shown), the overall agreement was 88% (95%CI=84–92%) and the kappa value was 0.71 (95%CI=0.62–0.72), with results from HPV testing of the self-collected specimens being more likely to categorize women in a higher HPV-risk category (p=0.009) (data not shown). Among WLWH, the overall agreement was 84% (95%CI=81–87%) and the kappa value was 0.76 (95%CI=0.74–0.77), with results from HPV testing of the self-collected specimens being more likely to categorize women in a higher HPV risk category (p=0.001) (data not shown).

Discussion

In this pilot study of cervical human papillomavirus DNA detection in WLWH and HIV uninfected women living in Limbé, Cameroon, the first study in this coastal town of Cameroon, we found an HIV prevalence rate 4.5% among women with previously unknown HIV status. We also found that crude HPV DNA prevalence was higher among WLWH compared to HIV[−] women. We also found older age and higher CD4 cell count were also more likely to be associated with lower HPV DNA prevalence and finally a good-to-excellent agreement for HPV detection between specimens.

Previous studies have reported a wide range of HPV prevalence in the inland regions of Cameroon. Untiet et al.⁷ reported a HPV prevalence (as detected by the Abbott Real-Time High-Risk HPV Assay) of 12.7% in provider-collected cervical specimens and 14.6% in self-collected cervicovaginal specimens in 649 women living in Yaoundé. Catarino et al.⁵ reported HPV prevalences (as detected by the cobas HPV test) of 35.8% in 547 HIV[−] and 66.7% in 60 WLWH aged 25–65 years living in Yaoundé and Edea. (Cohen’s kappa 0.74). Knunckler et al.⁴ reported a HPV and HPV16 prevalences (as detected by the Xpert HPV) of 18.5% and 1.8% in 1,012 women aged 30–49 years living in Dschang.

By comparison, we found i) the HPV, HPV16 and HPV 18/45 prevalences among HIV[−] were greater than that reported by Knunckler et al.⁴ (21.7% vs. 18.5%, 2.7% vs. 1.8%, and 4.8% vs. 2.9% on provider-collected specimens, respectively) and HPV prevalence was greater than reported by Untiet et al.⁷, (21.7% vs. 12.7% on provider-collected specimens); and 2) the HPV prevalence among WLWH was also greater slightly lower than that reported by Catarino et al.⁵ (46.4% vs. 39.0%, respectively). HPV16 is the most carcinogenic HPV genotype, accounting for approximately 55% of cervical cancers, and HPV18/45 are the next most important HPV genotypes, accounting for approximately 20% of cervical cancers, globally.¹¹

HPV prevalence in WLWH reported in this study was lower than observed in some populations^13;16;17, comparable to some populations^15;18;19, and higher than in other populations^14;20 living in SSA. However, some differences in HPV prevalence between studies may have been the result of age differences in study populations for which we could not account. Small differences in HPV prevalences between studies may have been introduced by using different HPV tests although of note Xpert benchmarks very well with other validated HPV tests^21–23 and has good reliability.^21,24 We observed, as seen in other studies, that HIV status^18;20, and CD4 counts among WLWH^15;20,25 were independent predictors of HPV prevalence.

The implication of the inverse relationship of CD4 counts and HPV prevalence are that good HIV control using antiretroviral therapy (ART) will reduce HPV prevalence, which on a population level is correlated with cervical cancer incidence.²⁶ An analysis from a cohort consortium of North American sites found an inverse relationship of CD4 counts and the risk of cervical cancer²⁷, which suggests that that HIV control will have significant population impact on cervical-cancer incidence. ART use reduces the risk of HPV persistence and progression to high-grade cervical abnormalities and cancer, and increases the likelihood of regression of cervical abnormalities.²⁸ One important implication of these and other data are 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 SSA where there is limited clinical infrastructure to manage and treatment those WLWH at risk for cervical cancer.

Perhaps surprisingly, we did not find a significant trend of older age groups with lower HPV prevalence among HIV[−] women that is commonly observed in many populations¹¹, as would be expected for any sexually transmitted infection. Notably, the sample size of HIV[−] women was small and only 39 women were aged 25–29 years, in whom the HPV prevalence was low. In a post-hoc analysis that excluded this small group of 25–29-year olds, the trend of older age group and lower HPV prevalence approached significance (p_trend=0.11) as did the trend when age was treated as a continuous variable (p=0.06). The inference is that there might have been a typical trend of older age and lower HPV prevalence but for a chance small sample of women aged 25–29 years with low HPV prevalence.

Finally, we observed good concordance for HPV detection between provider-collected and self-collected specimens, with the self-collection specimens more likely to test HPV positive than the provider-collected specimens. Similar observations were reported by Untiet et al.⁷ Small differences in the HPV positivity might have been introduced by the different sampling devices for the provider-collected sampling of cervix (Cervex Brush) and the self-collected sampling of the vagina (Viba Brush). However, we note that a recent meta-analysis found that when using a PCR-based method there was little impact of self-collection device on the relative performance of self-collection vs. provider-collection (uncontrolled for device) for detection of high-grade cervical abnormalities and cancer.²⁹ The inference is that sampling devices likely do not have a large influence on the detection of HPV.

One interpretation of these data is that cervical screening using self-collected specimens and HPV testing might be less specific than using physician-collected specimens in Cameroon, which is contrary to the results of a large meta-analysis that reported similar HPV positivity between self- and provider-collected specimens.²⁹ This difference in HPV positivity between specimens was observed in both HIV[−] and WLWH populations. In contrast, other studies in SSA of HPV testing have observed similar HPV positivity in paired self- and provider collected specimens.^30–34

This discrepancy in HPV positivity between self- and provider collected specimens was observed in this study and another study⁶ also conducted in Cameroon. Speculatively, women may have not inserted the self-collection device to the fornix and only sampled the epithelial tissue of the lower vagina, which independently can support a productive HPV infection.^35–37 This possibility is supported by another post-hoc analysis in which women were less likely to have a visible, acetowhite lesion on VIA (p=0.2) and abnormal cytology (from the provider-collected specimens) (p=0.005), both proxies of an HPV infection at the cervix, if the provider-collected specimen was HPV negative and the self-collected specimen was HPV positive than the converse. The causes of this discordance warrant further investigation.

The study population consisted of only women who sought care at the Limbé clinic and therefore may not be representative of the general population. This convenience sample excludes women who do not present at clinics for a variety of reason, are not HIV positive or on ART treatment. Therefore, this study population may have different and quite possible lower carriage of HPV infection than those not included in this study. This is a limitation of this study and the interpretation of our results warrant caution.

Highlights.

• There are limited data on cervical HPV prevalence in Cameroon and none from its Anglophone region. We investigated cervical HPV prevalence in HIV-uninfected (HIV[−]) and HIV-infected (WLWH) women living in the region. Study was conducted in Cameroon among a convenience sample 295 HIV negative women and 560 women living with HIV

• Crude HPV prevalence was 22% for HIV[−] women and 47% for WLWH (p<0.001). Among WLWH, older age and higher CD4 counts were associated with lower HPV prevalence. There was a good-to-excellent agreement for HPV detection between specimens.

• Self-collected were more likely than provider-collected specimens to test HPV positive, for all women and stratified by HIV status.

• HIV-related immunosuppression was a risk factor for HPV prevalence in this population and HPV testing of self-collected specimens appeared to be less specific than HPV testing of provider-collected specimens.