Cervical cancer (CC) is the second most common female cancer in women aged 15–44 years in the world, with its magnitude of difference in incidence by geography often striking (1). Sub-Saharan Africa (SSA) bears the highest burden of CC. Worldwide data has consistently reported that genital HPV infection is the most common sexually transmitted infection (STI) in most populations (1, 2), and that persistence of high-risk HPV (HR-HPV) infection causes anogenital cancers (2)—including CC (3). There is extensive overlap between geographical regions significantly burdened with CC and genital HPV infection. Epidemiological studies have found that HPV infection exhibits non-uniform geographical distribution, with SSA having the highest global prevalence (country prevalence ranges from 8.5 to 74.6% among women with normal cervical cytology) (1, 2, 4). Studies have noted a link between HPV infection and the age of women (1, 2, 4). These studies have observed that, whatever the geographical region and period of investigation, the early peak of HPV infection occurs at adolescent and young adult ages (<25 years). Thereafter, HPV prevalence declines with increasing age as observed globally (1, 2). However, in Africa, a second peak, though less pronounced, occurs in older women (≥45 years) (1, 2).

Globally, adolescent girls and young women (AGYW) remain a key sub-population most vulnerable to STIs, including HIV and HPV (4–13). Of note, a study that collated data from various sources in the U.S. found that 48.1% of the reported STIs occurred among youth (aged 15–24 years), of which at least half (50.5%) had incident HPV infections (14). A cross-sectional, multicentric, nationwide Brazilian survey found that 54.6 and 38.6% of sexually active unvaccinated AGYW had HPV and HR-HPV infections, respectively (8). Among indigenous Panamanian AGYW learners, HPV prevalence was found to be 33.2% (10). It has been noted that, among AGYW, the cumulative prevalence of HPV infection can exceed 80%, and that persistent HPV infection, to a lesser extent, is associated with cervical cytologic abnormalities (11). HPV infections among AGYW in SSA are significantly higher (4–6, 12), including in rural settings (12), than in the general populace of SSA and the rest of the globe (1, 2). South Africa still has scarce data on HPV infection among AGYW. The few published studies have reported that HPV and HR-HPV prevalence among AGYW can be as high as over 70 and 50% (4, 6, 12), respectively. These observations are confirmed and reinforced by the current high prevalence of HPV (76.1%) and HR-HPV (54.5%) AGYW from the Eastern Cape Province of South Africa (15). Published data on South African AGYW have also reported that multiple HPV infection varies by geographical area (5). In these HPV studies on South African cohort of AGYW, the determinants of HPV infection remain inconsistent and largely unexplored.

The determinants of HPV infection among women, including AGYW, appear to be very similar. The early peak in HPV prevalence often coincides with age at first sexual intercourse, which occurs during adolescent and young adult ages for most women. Both adolescent age and sexual debut in adolescence have been linked to prevalent HPV (16) and incident HR-HPV infection (17). Epidemiological studies have convincingly shown that HPV infection is rare before the first coitus (7, 13, 18, 19), and that HPV transmission in virgins is partly because of non-penetrative sexual contact (13) or douching habits (7). The rapid acquisition of HPV among AGYW after first coitus emphasizes the importance of vaccinating young girls and adolescents against HPV prior to early sexual debut. Other common risk factors (such as condomless vaginal sex, douching habits, and STIs) for HPV infection have been acknowledged (5, 12, 13, 16, 17, 19–21). Aside from these factors, vaginal microecology is gaining prominence on its role on HPV infection. In a meta-analysis, Candida albicans and bacterial vaginosis (BV, a vaginal syndrome-induced by an overgrowth of anaerobic bacteria) were identified as protective and risk factors for HPV infection, respectively (21). BV-associated bacteria (e.g., Gardnerella vaginalis) and vaginal microflora with paucity of keystone Lactobacillus spp. (e.g., Lactobacillus cripatus) are considered less protective against HPV infection (22–24). However, studies of the association of HPV (including HR-HPV) infection with vaginal microecology, specifically Candida spp. (20, 25) and Lactobacillus spp. (22, 23, 26), have yielded inconsistent and conflicting results. Hitherto, there are conflicting reports over the relationship of the relative abundance or dominance of L. iners with HPV infection; non-significant (24), positive (22, 27), and inverse associations (26) have all been reported. This could be ascribed to inter-study heterogeneity and clonal lineages of L. iners. Vaginal microecology and HPV have been associated with cervical intraepithelial neoplasia (21). Thus, more studies are needed to evaluate and understand the relationship between vaginal microecology and HPV infection, if we are to identify modifiable risk factors for HPV and CC prevention.

In Southern Africa, CC remains the most common cancer affecting women aged 15 to 44 years (1). Prophylactic HPV vaccines have been licensed in several countries as a primary preventive strategy for cervical HPV disease. Currently, two HPV vaccines are licensed in South Africa: (i) Cervarix®–a bivalent vaccine targeting two HR-HPV types (HPV16 and HPV18, responsible for majority of cancer, including CC), and (ii) Gardasil®4—a quadrivalent vaccine targeting HPV16, HPV18, and two low-risk HPV types (HPV6 and HPV11, the most common causative agents of genital warts) (28). Gardasil®4, administered in three doses (0, 1–2, 6 month schedule), was successfully used in an HPV vaccination demonstration programme in South Africa targeting girls aged 9–12 years (29). Nevertheless, in April 2014, South Africa's national Department of Health introduced a school-based HPV vaccination programme; with the administration of a two-dose regimen (6 months apart schedule) of the Cervarix®, primarily targeting girls aged ≥9 years in their fourth year of elementary (primary) school (Grade 4, equivalent to 4th Grade in the U.S. and Year 5 in the UK) (30). To realize the greatest impact of the HPV vaccination programme, the World Health Organization (WHO) recommends that the vaccines be administered to young girls (9–13 years) prior to sexual debut, hence before exposure to HPV infection (28). Most AGYW in South Africa do not have access to the HPV vaccine programme since only girls in Grade 4 in some public schools are prioritized (31). For AGYW aged 15–25 years who may not benefit from catch-up vaccination programmes, an understanding of the modifiable risk factors may help to reduce the duration of HPV infection. To address this, information on HPV epidemiology among AGYW need to be collected.

We are currently aware of the high HPV prevalence, including the genotypes targeted by Gardasil®9 (a non-avalent HPV vaccine) among unvaccinated AGYW in rural Eastern Cape (15), yet HPV risk factors among AGYW remain unknown. Thus, the aim of the present study was to determine the factors associated with genital HPV infection among AGYW in the Eastern Cape. This study may help formulate and implement evidence-based policies designed to create and intensify awareness of risk factors for HPV infection among AGYW in SSA.

STIs, including HPV infection, in AGYW is a pressing global health problem (6, 9, 10); resulting in significantly reduced quality of life of infected individuals. In South Africa, there is little to no available regional and nationwide data on the factors associated with HPV infection among AGYW. Therefore, in the present cross-sectional study, we assessed sociodemographic, behavioral, clinical, and microbiological factors associated with the highly prevalent (76%) HPV infection among AGYW in the rural Eastern Cape.

We identified a wide variety of STIs and pathobionts. The prevalence of any bacterial STI (34%), N. gonorrhea (7%), and C. trachomatis (30%) were intermediate to those reported among South African AGYW from Cape Town (60, 14, and 43%, respectively) and Johannesburg (19, 5, and 17%, respectively) (35). Our data add to the consensus that, after HPV infection, C. trachomatis is the second most prevalent STI among AGYW in South Africa (5, 6, 9, 35) and other regions (10, 14). Syphilis, chancroid, and LGV, were not detected whereas T. vaginalis, M. genitalium, HIV, HSV-1, HSV-2, and candidiasis were detected at low rates (1–10%), comparable to AGYW cohorts from different regions in South Africa (5, 9, 35). Overall, the distribution pattern of viral and bacterial STIs and pathobionts among the AGYW resembles that previously reported among Eastern Cape women aged ≥30 years attending community-based clinics or CC screenings (36). However, the prevalences of these microbes, including infections, are higher among AGYW compared to the aforesaid Eastern Cape women (37); thus, reflecting high-risk sexual behavior among AGYW. We also affirmed that BV prevalence (44%) is high among South African AGYW, and regionally ranges from 36 to 51% (35). We find the presence of a wide range of STIs among the HPV-burdened unvaccinated AGYW learners worrying, as it may compromise HPV prevention and control. Studies have already found a positive relationship of HPV with other STIs (20, 21, 36). A small cohort study on U.S. adolescents noted that HPV infections were acquired earlier than other STIs (18). This probably corresponds to the early peak of HPV infection. If indeed HPV infections among AGYW precede most of the other STIs, then it makes sense to ensure that, as many young girls as possible are vaccinated against HPV and educated about human reproductive health prior to their sexual debuts. Perhaps these may substantially reduce the burden of HPV and other STIs.

Owing to the high prevalence of HPV infection in our study cohort (15), current and projected (2030) high caseload of CC in Southern Africa (1, 2), variations of HPV risk factors by community (geography) (5, 8, 12, 16), and recent reporting of HPV-related cervical cytologic abnormalities among women aged ≥18 years in rural Eastern Cape (38), we sought to evaluate the determinants of HPV infection among these AGYW. In univariable analysis, alcohol consumption, candidiasis, U. urealyticum, BV and specific BV-associated bacteria were strong predictors of HPV infection. We also found that besides alcohol consumption, vaginal discharge or itching was mainly associated with specific vaginal microbes, all of which were associated with HPV infection. A previous study on Swedish cohort aged 17–50 years noted that vaginal discharge and itching were more common in women with vulvovaginal HPV infection (39). However, our study found no direct association between vaginal HPV infection and vaginal discharge or itching. Abnormal vaginal discharge and itching are probably because of vaginal infections or syndromes, including BV, an independent predictor of HPV.

Alcohol consumption is a common behavior associated with HPV infection (16). It may lead to irrational choices, including high-risk sexual activities (e.g., engaging in condomless vaginal sex) that increase the risk for STIs (40). A survey that estimated alcohol-related health problems among U.S. college students (aged 18–24 years) had 8.4% of the respondents attributing the cause of engaging condomless vaginal sex to alcohol consumption (41). However, in our study, condom use was not a determinant of HPV infection, yet there was a trend toward association of both the frequency of alcohol consumption and consumption of different alcoholic drinks in the past month with HPV infection. This could have been influenced by biased self-reporting of condom use, particularly on events that occurred under the influence of alcohol. We know that alcohol consumption may cause memory impairment and blackouts (42). In congruent with other public health experts (43, 44), our findings suggest that didactic public health campaigns advocating for the reduction of high-risk sexual behavior among AGYW are warranted to curtail HPV infection. School-based interventions with risk-modification counseling can control HPV infection, for example, by improving HPV knowledge and awareness (32), intention to use condoms, and uptake of HPV vaccine among adolescents (45). Currently, there is noticeable poor knowledge of HPV, its associated risk factors, and prevention among AGYW (43, 44), including in our study population (32) and other parts of SSA (31).

There is still a gap between the relationship between vaginal microecology (fungi and bacteria in our context) and HPV infection. There have been mixed findings with regards to the association of candidiasis with HPV and HR-HPV infections (20, 21, 25). We found that, in contrast to a previous report (21), C. albicans significantly predicted HPV infection. The observed differences probably reflect methodological and true population differences, which include variations in host epithelial cells (46) and failure to resolve dimorphic switches of C. albicans (47). Under normal scenario, Candida spp. are human commensal microflora that live in harmony with normal microflora. We think that their ability to form biofilm matrix partially account for their inverse association with HPV infection. This is because the dispersed form of C. albicans is more virulent than mature C. albicans biofilms (47). C. albicans positive association with HPV is, perhaps contributed by adhesion and cellular damage to mucosal epithelia (46, 47). A breeched epithelial barrier may then act as a gateway for HPV and replication in the basal layer (48).

The associations of specific BV-associated bacteria, BV, and pathobionts with HPV infection are in common with other reports (21, 23). Serving as an example, Mbulawa et al. (5) reported an association between BV and HPV among AGYW from Cape Town and Johannesburg in South Africa. These findings are not surprising as a majority of women from SSA, including AGYW from South Africa, have cervicovaginal microflora devoid of appreciable quantities of the protective lactobacilli (35, 49). Furthermore, vaginal microflora with higher abundances of keystone lactobacilli are thought to be protective against HPV (HR-HPV) infection whereas those with BV-associated bacteria (including A. vaginae) or BV are not (10, 22). In the only study to date of cervicovaginal microflora and HPV infection in South Africa, higher relative abundances of A. vaginae and G. vaginalis [believed to initiate BV development (50)] were correlated with HPV infection (including HR-HPV) (49). A. vaginae and sialidase gene from G. vaginalis have been associated with persistent HPV infection (27). The mechanisms that BV-associated bacteria use to contribute to HPV infection partly parallels that of C. albicans. An in vitro model found that besides BV-associated G. vaginalis manifesting propensity to displace protective lactobacilli from HeLa epithelial cells, caused cytopathogenic changes on these epithelial cells (50). Cervicovaginal microflora with G. vaginalis dominance has been associated with host epithelial barrier disruption and impaired wound healing (51). As earlier explained, HPV infection of the basal layer requires disruption of epithelial barriers.

With regards to pathobionts, U. urealyticum, has been associated with both increased HPV infection (21) and young women age (≤25 years) (52). It is possible that U. urealyticum is also co-acquired during the early peak of HPV infection. Also, this assumption is reinforced by the finding that U. urealyticum is more frequent in women with BV than healthy women (53) and the possibility of heterosexual exchange of BV-associated bacteria (54). Ureaplasma/Mycoplasma infections may be risk factors for persistent HR-HPV infection and could be associated with the development of precancerous cervical lesions (55). U. urealyticum could be causing pathogenicity by synergistic effects with other pathobionts and BV associated bacteria. This theory stems from the association of U. urealyticum/U. parvum coinfection with incident C. trachomatis (52) and the perspective that U. urealyticum might have better survival in symbiotic relationships with BV-associated bacteria (53). In-depth studies are required to delineate the role of U. urealyticum and other pathobionts in HPV infection. Since these pathogens are ubiquitously present in the urogenital tract in both healthy and diseased individuals (52, 53), their quantitative and not only qualitative molecular detection should be considered in order to improve their predictive values (56). Limited knowledge of the relationship between vaginal microecology and HPV infection remains a key “Achilles heel” of HPV control programmes. Significant associations of specific vaginal microecology with HPV infection underline the value to screen for and manage aberrant vaginal microflora (i.e., those with bacteria indicative of BV) within an integrated model for HPV control among AGYW.

The major strengths of our study are the rich participant information and the utility of a culture-independent approach for the detection and/or quantification of vital organisms in the vaginal microflora. Molecular-based methods perform better than the culture method, which, for instance, cannot discriminate between U. urealyticum and U. parvum. Consequently, such culture results are inaccurately reported as U. urealyticum instead of ureaplasmas (56). Our findings should be considered in light of the following study limitations. In addition to its cross-sectional nature, our study included only AGYW from two schools (recruited from convenient primary care facilities) in the rural Eastern Cape, therefore may have limited its casual inferences and generalizability. Since the vaginal specimens were self-collected by inserting the brush as far as possible into the vagina after assuming a comfortable position, inter-individual differences (inconsistent sampling) in anatomical site sampled cannot be ruled out. This may have affected our results since the vaginal milieu comprises different anatomical sites, which are composed of varying microbial community composition (57). Another study limitation is the inability of the multiplex real-time PCR assay to distinguish between bacterial colonization and pathogenic infection when examining the pathobionts. We do not rule out the possibility of bias since we relied on self-reported questionnaire. Finally, we did not adjust for possible confounders for HPV infection, including sexual networks. Although not investigated, the observed HPV prevalence and its factors associated with HPV infection could somewhat be mediated by (i) poor HPV vaccine acceptance and coverage (30, 31), and (ii) poor community knowledge of HPV infection, prevention, and its consequences (31, 32, 43, 44). Notwithstanding the study limitations, our study leads us to believe that HPV education should be urgently integrated in the high school educational curriculum to confront the high HPV burden among AGYW in SSA.

Our study is the first to evaluate the factors associated with HPV infection among AGYW learners in the Eastern Cape Province of South Africa. Our study, albeit not necessarily generalizable, yet consistent with published literature from other female sub-populations, found specific factors (including vaginal microecology) associated with prevalent HPV infection among unvaccinated AGYW in South Africa. The association of HPV infection and vaginal discharge or itching with specific microbes in the vaginal microflora underlines the necessity of incorporating such microbes in the surveillance and management of vaginal syndromes and infections, including HPV infection among AGYW. But first, further surveillances are needed to: (i) accurately investigate HPV epidemiology and its determinants in other AGYW populations, and (ii) adequately resolve their roles in HPV infection. To conclude, the findings from our study have the potential to inform comprehensive evidence-based interventions and programmes which, increase the public health awareness of HPV infection, its risk factors, and ramifications among AGYW in SSA.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The parent study – The HPV Education Intervention Study – was reviewed and approved by the Human Research Ethics Committee at the University of Cape Town, South Africa (HREC: 369/2015) and the Eastern Cape Province of South Africa was granted by the Eastern Cape Provincial Health Research Committee (EC_2016RP29_562). Moreover, Eastern Cape's Department of Education and Department of Health also granted us permission to conduct the study. Written informed consent to participate in parent study was provided by the participants' legal guardian/next of kin. For the present study, which sought to include a molecular test of a panel of vaginal microbes, ethical clearance was secured from the Human Research Ethics Committee at Walter Sisulu University, South Africa (protocol number: 044/2020).

A-LW and ZM: conceptualization and resources. ZM: methodology, validation, supervision, project administration, and funding acquisition. HO and SM: formal analysis. HO, SM, A-LW, and ZM: investigation and writing—review and editing. HO, SM, and ZM: data curation. HO: writing—original draft preparation and visualization. All authors contributed to the article and approved the submitted version.

The present work was funded by research grants of the South African Medical Research Council (SIR Grant Number: 384709 and SAMRC/UCT Gynaecological Cancer Research Centre) and the National Health Laboratory Service Research Trust. HO was funded by the University of Cape Town's Faculty of Health Sciences Postdoctoral Research Fellowship and the South African National Research Foundation Postdoctoral Grantholder Bursary (Grant Number: 64815).

Author HO is employed by Tunacare Services Health Providers Limited, Nairobi, Kenya. The remaining authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

The findings, views, conclusions, and recommendations expressed in this article are those of the authors and do not necessarily represent the views of the funders and institutes affiliated with the authors and/or those mentioned in the “Acknowledgments” section above.