Relationship between Highly Active Antiretroviral Therapy (HAART) and human papillomavirus type 16 (HPV 16) infection among women in Sub-Saharan Africa and public health implications: A systematic review

Invasive cervical cancer is the most prevalent cancer among women in Sub-Saharan Africa. In 2013, the World Health Organization (WHO) emitted recommendations to start Highly Active Antiretroviral Therapy (HAART) regardless of CD4 count. Although HAART has been shown to reduce the prevalence of high-risk human papillomavirus (HR-HPV) genotypes, it is unclear whether it confers a protective effect specifically for HPV 16. This review summarizes the existing evidence regarding the effect of HAART on HPV 16 infection, as this genotype may not be influenced by immunity level and explores its implications for Sub Saharan Africa. A comprehensive literature review was undertaken and quality assessment was carried out on the selected papers. Four cohort studies and three cross-sectional studies were identified for which the overall quality score assessment ranged from weak/moderate (Score of 1.8) to strong (Score of 3). The evidence yielded by our review was conflicting. Thus, the high heterogeneity between study populations and results did not allow us to draw any firm conclusions as to whether HAART has an impact on HPV 16 acquisition/prevalence. As only three studies were conducted in Africa, there are insufficient grounds for solid comparison between geographic regions. In light of inadequate data, HPV unvaccinated women on HAART should still receive more frequent follow-up.

Introduction 19]. Some studies have observed a higher relative prevalence of non-HPV 16 genotypes in severely immunocompromised women as well as in women with no access to HAART [19][20][21]. This may be attributed to a poor immune response against non-HPV 16 genotypes in severely immunosuppressed HIV infected women [18,20]. It can be hypothesized that the variable effect of HAART in HIV infected women in Africa is contingent on the prevalence of HPV type 16 [22], which improved immunocompetence may not prevent. However, epidemiological evidence of this is still scattered and inconclusive. Should epidemiological evidence confirm that HAART has no impact on reducing HPV 16 prevalence, the current WHO recommendation of not rescreening women for at least 5 years, but within ten [23], among those screened negative will need to be revisited.
The primary objective of this review is to elucidate the impact and effects of HAART on prevalence/incidence of HPV 16 compared to other genotypes. The second objective is to highlight knowledge gaps for further research and specifically how they relate to the sub-Saharan African context.

Search strategy and selection criteria
We conducted this systematic review based on a pre-defined search protocol that conformed to the criteria set out by the PRISMA statement. PubMed, CINAHL, Global Health, Scopus, and EMBASE databases were searched between January 1997 and March 9, 2018 (S1 File).

Search terms
The domains of the search terms were HPV 16 OR human papillomavirus genotype 16 AND acquisition OR incidence/detection OR prevalence AND Anti-HIV Agents OR Antiretroviral Therapy OR High Active OR antiretroviral OR ART OR HAART (S3 File)

Study selection
We considered the following four components (Population, Intervention, Comparison, Outcome-PICO) to assess and categorize studies to be included in this review. The population involves HIV-positive women with HPV 16 infection (among other genotypes). The intervention examined in this study is the administration of HAART, defined as at least 3 antiretroviral drugs belonging to 2 different drug classes (Non-nucleoside reverse transcriptase inhibitors or nucleoside reverse transcriptase inhibitors or Protease Inhibitors) and, if possible, the duration of treatment. Women on HAART for at least 6 months are being compared to women not receiving HAART or just starting HAART. The outcomes of interest are: Prevalence or incidence or acquisition of HPV 16 for the two comparison groups.
We initially intended to conduct a systematic review and meta-analysis to answer the question on whether HIV positive women on HAART have higher probability to prevent HPV 16 infection than HIV-positive non-users of HAART. However, due to the scarcity and methodological heterogeneity of retrieved studies, and the lack of similarity of measure of effect of the various studies (Odds ratio vs Hazard ratio), pooling estimates would be inappropriate. We therefore re-oriented this study to a systematic review, without conducting a meta-analysis. The search strategy was designed by a medical librarian to identify studies reporting HPV 16 acquisition in women on HAART. We conducted this review on a pre-defined search protocol that conformed to the criteria set out by the PRISMA (S2 File) statement [24,25]. (S2 and S3 Files)

Data abstraction
All studies were independently reviewed and critically evaluated for inclusion by three authors (SM, ZN and RR). All data were extracted independently and in duplicate manner by the three investigators (SM, ZN and RR). Possible cases of disagreement were resolved on an individual basis via detailed discussion among the authors of the justification for exclusion/inclusion. The following items were recorded: first author, publication year, study design, type of sampling, sample collection method, study population, total sample size, HIV status, incidence/ prevalence of HPV 16 and duration of HAART use, baseline of CD4 count in study population, sexual behavior, prevalence of multiple HPV genotypes, past history of STI and if the analysis included adjusting for confounding factors. The adjusted Odds ratio (a)OR, adjusted Hazard Ratio(a)HR and 95% Confidence interval (CI) were recorded and if the 95% CI did not include 1 and no p-value was given, it was considered significant.

Data assessment
Because of the lack of randomized trials retrieved (due to ethical considerations), we considered prospective cohort/longitudinal studies to be of highest quality. We reported an overall score (from 1 to 3) based on our assessment of the overall quality of evidence for the following parameters: 4. Withdrawals, loss to follow-up and dropouts: (Score "3" = strong), description of a strategy to minimise the loss to follow-up, reported withdrawals and drop-outs (80-100% of participants completed the study) [27]; (Score "2" = moderate), reported withdrawals and dropouts (60% of participants completed the study); (Score "1" = weak), withdrawal and dropout rates not specified.
We assessed the methodological quality of evidence by calculating the mean of each study and attributing the mean one of the qualitative attributes, strong, moderate and weak.

Ethical approval
No ethical approval was required as this was a systematic review (analysis of secondary aggregated data).

Search results
On March 8, 2018, we retrieved 210 articles from PUBMED, CINAHL, Global Health, Scopus and EMBASE, of which 103 duplicate titles were identified. After screening the 107 remaining titles and abstracts, 11 were full text-screened and 7 were eventually included in this review.
No disagreement between the three reviewers emerged on whether to include or exclude any given study.

Study population
Studies varied in reporting, baseline of CD4 count in study population, sexual behavior, prevalence of multiple HPV genotypes, and past history of STI (Table 1).

Study design
A total of 3298 HIV infected women were included in the studies retrieved for this review. Sample sizes for these studies ranged from 225 to 1010 HIV infected women.
Seven studies were eventually included in this review. These studies varied by their design including prospective (n = 4) and retrospective cohort (n = 1) and cross-sectional (n = 2) studies. Four studies were conducted in developing countries (3 in sub-Saharan African countries and 1 in India) while the other three were in developed countries (USA, Canada and Belgium). Table 2 Quality assessment of studies The overall quality of evidence pertaining to the association/impact of HAART on HPV 16 prevalence/incidence ranged from weak/moderate (Score of 1.8) to strong (Score of 3). One study had a score of 1.8 [28,29] four studies had an average score between 2 and 3 [8,[29][30][31], hence the quality of their evidence was considered as "moderate to strong" and one study had a strong score of 3 [32].The weak ratings mainly resulted from lack of control of confounding factors (Table 3)

Systematic review results
HAART and HPV 16 prevalence in comparison with other HPV genotypes. All seven studies explored the impact of HAART on HPV 16 prevalence or incidence, of which four supported our hypothesis that HAART has no impact on reducing HPV 16 acquisition or prevalence. In a longitudinal cohort study of 750 HIV-infected women, Blitz (Canada, 2013) [8] found that HAART did not significantly reduce the acquisition of HPV 16 (univariate HR: 0.52, 95% CI: 0.15-1.81) or non-HPV 16/18 oncogenic HR-HPV types (univariate HR: 0.94, 95%CI: 0.41-2.16). In a multivariate analysis, (Belgium, 2013) [29] at same median duration on ART, the prevalence of HPV 16 was not significantly different from the prevalence of other HR-HPV, 90% versus 86%, respectively.
We identified three studies, which refute our hypothesis. Zeier (South Africa, 2015) [32], in the univariate analysis, reported that HAART in a univariate model for detection of any HPV genotype, and HAART status (receiving HAART or not) as the only predictor, HAART reduced the risk of HPV presence by 77% (OR: 0.23, 95% CI: 0.15-0.37). HAART significantly reduced the detection risk for cervical HR-HPV presence (OR: 0.33, 95% CI: 0.24-0.44) and HPV-16 (OR: 0.50, 95% CI: 0.37-0.67). In the adjusted model, longer duration on HAART Relationship between incidence/prevalence of HPV 16 and HAART, systematic review

Discussion
The findings from this review are conflicting and contradictory. Therefore, a conclusive answer on the impact and effect of HAART on the incidence and prevalence of HPV 16 is not possible. The overall quality of evidence pertaining to the association/impact of HAART on HPV 16 prevalence/incidence ranged from weak/moderate to strong, which further reduces our capacity to explore the association. Our study builds on the recent global meta-analysis [17], which explored the pooled effect of HAART on aggregate HR-HPV prevalence, by identifying other studies which exclusively focused on comparing incidence or prevalence of the most virulent, carcinogenic and aggressive HPV genotype (HPV 16) to other HR-HPV genotypes. Moreover, given the fact that only three studies took place in Sub Saharan Africa, we are limited in our capability to compare findings between this geographical region and others. This was exacerbated by the high degree of heterogeneity in quality, reporting, setting, study population and results. The baseline CD4 count was mostly given upon enrollment of in the study, but not when HAART was initiated. And in only one study [30], the median CD4 count at baseline was known in HPV infected women on HAART with HPV 16.
Furthermore, only one study examined past history of STIs [33]. A meta-analysis suggested a significant association between bacterial vaginosis (BV) and HR-HPV genotypes [34]. HPV 16 and its phylogenetically-related genotypes have also been linked with BV. Studies have suggested a significant association between BV and HPV 16 in female sex workers in Spain [35], and a marginal association between BV and HPV 58, a HPV 16-related type in Kenya [36].
Moreover, only one study [8] examined regular condom use in the past six months and recent sexual partners, which may be more relevant confounders to adjust for than the number of past sexual partners, as previous HPV 16 infection may have cleared up in the interim. Due to variations in intervention duration, sample size, study location, individual variations in CD4 counts and studied the populations studied, evidence on the impact of HAART and HPV acquisition in HIV-infected women remains inconclusive. No study examined HIV viral load during HAART treatment.

Strengths and limitations of this study
The major strength of this study is that there has been a thorough attempt to search five databases and retrieve all the relevant existing literature. Nonetheless, this study does have limitations as well, including our incapacity to carry out a meta-analysis due to the heterogeneity of study designs and their measurement of effect.

Public health and clinical implication
In light of the inconclusive findings of the review, unvaccinated HIV-infected women may still require more regular monitoring than once every three years if tested negative by VIA or cytology, as recommended by the WHO for this population [37], despite recent findings that HAART may reduce HR-HPV prevalence [17]. In turn, these findings underscore the importance of primary prevention of HPV-16 acquisition through prophylactic vaccines due to the scarcity of evidence on whether HAART is specifically able to confer a protective effect against HPV 16 as it does against other HR-HPV genotypes [17].

Research gaps
A major gap is to elucidate the HPV 16 Basic Reproductive Number (R0) in various types of populations, which may reveal what percentage of the population needs to be vaccinated. In light of the rollout of HAART at a higher CD4 count in sub-Saharan Africa, it is imperative that the prevalence of HPV 16 in HIV-infected women per sub-Saharan country be better established. Knowledge about the specific HPV type distribution in the HAART era is crucial to guide HPV-based screening intervals and the feasibility of catch-up campaigns focusing on older HIV-infected girls.
Recent literature suggests that the lesser impact expected of the HPV vaccine in older HIVinfected girls, due to a potential higher risk of prior exposure to vaccine-targeted genotypes, may be compensated by the potential of the HPV vaccine to prevent recurrence of CIN 2-3 among women treated for HSIL [38]. In this population, the incidence of disease recurrence after treatment varies widely from 25 to 55% at 12 months [39]. In addition, research is required to explore whether HAART initiation at a less immunocompromised state can prevent subsequent HPV 16 infections by generating antibodies and hence providing protection against new infections.
In light of the increasing number of HIV-infected women accessing HAART at a less immunocompromised state, the impact of different levels of immune reconstitution conferred by HAART on HPV 16 acquisition as a stand-alone or as co-infection with other HR-HPV genotypes should be elucidated. Hence, it is imperative that more studies on the sub-Saharan continent be undertaken and disaggregated baseline data be collected by HPV 16 status in order to adjust for potential confounders.
As the prevalence of BV in sub-Saharan women is among highest in the world [40], it is pivotal that the association between BV and HPV 16 in women on HAART is elucidated and whether it would be cost-effective to launch catch-up campaigns among HIV-infected women below 26 years of age.
Rollout of HAART in sub Saharan Africa has brought forth concomitant increases in drug resistance rates, which can be attributed to the empirical first and second line antiretroviral regimens and the clinical or immunological definitions of treatment failure without plasma viral load monitoring [12]. Data on second-line failure and development of PI resistance in sub-Saharan Africa are scarce [41]. As third line drug options, such as integrase inhibitors, including dolutegravir, are expected to increase [42], longitudinal studies are required to elucidate the impact of HAART on HR-HPV genotypes as well as on HPV 16 acquisition in HIVinfected women.

Conclusion
Our limited data does not enable us to yield conclusive findings. To err on the side of caution, our data suggests that unvaccinated women on HAART may require more frequent follow-up. The scarcity of data on acquisition of specific HPV genotypes by HAART regimens is reflective of the lack of data found on the immuno-epidemiology of oncogenic genotypes and their synergistic interactions prior and after initiation of HAART. Apart from recent literature suggesting a protective effect of HAART against HR-HPV infection, further research on the impact of HAART on HPV 16 infection is needed with a view to designing an optimal screening and vaccine strategy in HIV-infected women in Sub Saharan Africa.