The effectiveness of vaccination to prevent the papillomavirus infection: a systematic review and meta-analysis

Our purpose was to determine the effectiveness and harms of vaccination in patients with any sexual history to prevent the prevalence of papillomavirus infection. A search strategy was conducted in the MEDLINE, CENTRAL, EMBASE and LILACS databases. Searches were also conducted in other databases and unpublished literature. The risk of bias was evaluated with the Cochrane Collaboration's tool. Analysis of fixed effects was conducted. The primary outcome was the infection by any and each human papillomavirus (HPV) genotype, serious adverse effects and short-term adverse effects. The measure of the effect was the risk difference (RD) with a 95% confidence interval (CI). The planned interventions were bivalent vaccine/tetravalent/nonavalent vs. placebo/no intervention/other vaccines. We included 29 studies described in 35 publications. Bivalent HPV vaccine offers protection against HPV16 (RD −0.05, 95% CI −0.098 to −0.0032), HPV18 (RD −0.03, 95% CI −0.062 to −0.0004) and HPV16/18 genotypes (RD of −0.1, 95% CI −0.16 to −0.04). On the other side, tetravalent HPV vaccine offered protection against HPV6 (RD of −0.0500, 95% CI −0.0963 to −0.0230), HPV11 (RD −0.0198, 95% CI −0.0310 to −0.0085). Also, against HPV16 (RD of −0.0608, 95% CI −0.1126 to −0.0091) and HPV18 (RD of −0.0200, 95% CI −0.0408 to −0.0123). There was a reduction in the prevalence of HPV16, 18 and 16/18 genotypes when applying the bivalent vaccine, with no increase in adverse effects. Regarding the tetravalent vaccine, we found a reduction in the prevalence of HPV6, 11, 16 and 18 genotypes, with no increase in adverse effects.


Introduction
The development of the three FDA-approved multivalent prophylactic human papillomavirus (HPV) vaccines has been because of the discovery of HPV infection as the cause of all cervical cancer [1]. The three prophylactic HPV vaccines have shown high efficacy for prevention of HPV infection [2].
Cervarix® and Gardasil® were the first vaccines for the prevention of the cervical cancer. Cervarix® targets types 16 and 18 HPV, which are responsible for 70% of all cervical cancer [3], while Gardasil®, on the other hand, adds also activity against types 6 and 11 HPV, which cause 90% of anogenital warts [4]. Additionally to those included in the qHPV, the 9vHPV vaccine contains type 31, 33, 45, 52 and 58 antigens [5].
These vaccines are well-tolerated, safe and only with minor adverse effects. They also protect against pre-cancerous lesions caused by subtypes 16 and 18 in a naive population, adding a systemic immune response at 5 years post-vaccination [6][7][8].
The primary care service in United States recommended HPV vaccine for females aged 9-26, through the Vaccines for Children program since 2006 [9]. Additionally, the immunisation for adolescents aged 11-12 years and for adolescent women (aged 13-26 years) before sexually active was recommended by The US Center for Disease Control and it is accepted in many developed countries [7,10].
According to the health care system's organisation, the coverage rate and the programmes differ. Some of the vaccination programmes are offered through schools (e.g. Australia, UK) whereas others are provided in private clinics, or public primary care (e.g. the United States) [11,12].
Different studies (phases II and III) have been conducted, and a few reviews have tried to pooled the effects [13], however, none of them emphasise the changes in the prevalence/ incidence of HPV serotypes in the populations. Besides, previous reports have not assessed the impact of vaccination in different ages and time since sexual beginning, nor multisite and cross-protection against HPV types. Also, there is conflict about the protection against re-infection among individuals known to be infected and who subsequently clear their infections [14].
The objective was to determine the effectiveness and harms of vaccination in patients with any sexual history to prevent the prevalence of papillomavirus infection.

Methods
We performed this review according to the recommendations of the Cochrane Collaboration [15] and following the PRISMA Statement [16]. The PROSPERO registration number is CRD42017074007.

Eligibility criteria
We included only clinical trials; both gender patients from any population at any age and any sexual history. The interventions were: vaccine against HPV: HPV16 and HPV18 (bivalent vaccine), or HPV6, HPV11, HPV16 and HPV18 (tetravalent vaccine) and HPV6, HPV11, HPV16, HPV18, HPV31, HPV33, HPV45, HPV52 and HPV58 (nanovalent) genotypes. The vaccination had to be intramuscular injection over a period of 6 months according to the standard scheme. We did not include pregnant patients The comparisons were: bivalent vaccine/tetravalent/nanovalent vs. placebo/no intervention/other vaccines. Outcomes: • Infection by any HPV genotype: assessment must be by any validated technique identified by the blood sample and oral cell sample. If patients are older than 18 years old, the sample must come from the specific organ. • Infection by each HPV genotype: assessment must be by any validated technique identified by the blood sample and oral cell sample. If patients are older than 18 years old, the sample must come from the specific organ. • Serious adverse effects.
• Short-and long-term adverse effects.
For infection and long-term adverse effect outcomes, studies had at least 1-year follow-up and for short-term outcome studies were allowed to be less than a 1-year follow-up.

Information sources
A literature search was conducted as recommended by Cochrane. We used medical subject headings (MeSh), Emtree language, Decs and text words related. We searched MEDLINE (OVID), EMBASE, LILACS and the Cochrane Central Register of Controlled Trials (CENTRAL). The search was performed from 2000 to to date.
To ensure literature saturation, we scanned references from relevant articles identified through the search; conferences (HPV2015 and HPV2017) related to HPV and vaccines and thesis databases (e.g. Theseo). Grey literature was searched in open grey and Google scholar. We looked for clinical trials in the process in clinicaltrials.gov, the registry created by the World Health Organization and the New Zealand registry for Clinical trials, among others. We contacted authors by e-mail in case of missing information. There were no setting or language restrictions.

Data collection
Two researchers reviewed each reference by title and abstract and full-texts, applied pre-specified inclusion and exclusion criteria. Disagreements were resolved by consensus and where the dispute could not be solved, a third reviewer dissolved conflict.
Two trained reviewers using a standardised form independently extracted the following information from each article: study design, geographic location, authors names, title, objectives, inclusion and exclusion criteria. Also, the number of patients, type of laboratory technique, losses to follow-up, timing, definitions of outcomes, outcomes and association measures and funding source.

Risk of bias
We assessed the risk of bias for each study with the Cochrane Collaboration tool, which covers: sequence generation, allocation concealment, blinding, incomplete outcome data, selective reporting and other biases. Two independent researchers will judge about the possible risk of bias from extracted information, rated as 'high risk,' 'low risk' or 'unclear risk. ' We described a graphic representation of potential bias using RevMan 5.3 [17].

Data analysis/synthesis of results
We performed the statistical analysis in R [18]. For categorical outcomes, we reported information in risk differences (RD) with 95% confidence intervals (CIs) according to the type of variables, and we pooled the data with a random effect meta-analysis according to the heterogeneity expected. The results were reported in forest plots. Heterogeneity was evaluated using the I 2 test [15]. For the interpretation, it was determined that the values of less than 50% would be for low heterogeneity and more than 50% would be for the high level of heterogeneity.

Publication bias
We did not perform publication bias due to the number of included studies in each meta-analysis.

Sensitivity analysis
We performed sensitivity analysis extracting weighted studies, analysing the length of follow-up and running the estimated effect to find differences.

Subgroup analysis
We tried to perform subgroup analysis by co-infection, sample site, gender, age and continent however due to the scarcity of data, we were not able to do it.

Results
We found 2834 records through the electronic search strategy and 20 through other searches (Fig. 1). After excluding duplicates, we ultimately included 29 unique studies described in 35   The rest of the information regarding the included studies is described in Table 1.

Excluded studies
Sixty-eight full-text articles were excluded (28 studies did not have the outcome of interest; 17 were not RCT; 11 studies had no intervention of interest; four studies did not have enough follow-up; four had no population of interest, and four were not related).

Risk of bias assessment
We found that 50% and 75% of the included studies were graded as unclear on describing the random sequence generation No selective reporting was identified during grading in these studies. Nonetheless, all studies were industry-funded, and this leads to unclear classification (Fig. 2).
Regarding the HPV18 genotype, we found five studies (Konno, 2014   Other results for any genotype, little adverse effects and autoimmune disease are shown in Table 2.
Other results for little adverse effects and autoimmune disease are shown in Table 2.
Other results for HPV16/18 and small adverse effects are shown in Table 2.
Other results regarding short-term adverse effects and serious adverse effects are shown in Table 2.
Nonavalent vs. tetravalent: total Only one study (Vesikari 2015) compared the 9vHPV against the tetravalent vaccine. The study was conducted in different countries and included people from 9 to 15 years. The outcome was evaluated at 7 months and the results were: anti-HPV16 and anti-HPV18, geometric mean titer (GMT) were similar between vaccines (anti-HPV16 GMTs: 6739.5 vs. 6887.4 mMU/ml for

Summary of the main results
In summary, evidence suggested that the bivalent HPV vaccine offers protection against HPV16, HPV18 and HPV16/18 genotypes without increasing adverse effects. It is consistent with 4 years for HPV16 and HPV18 genotypes. On the other side, tetravalent HPV vaccine, offered protection against HPV6, HPV11, HPV16 and HPV18 without increasing adverse effects and this was consistent at 4 years.

Comparison with other reviews or other studies
Detecting a specific capsid antibody in serum suggests an HPV infection. Although more than 50% of infected subjects will have a serological conversion, and thus detection of HPV antibodies. This has been used to analyse the natural history and the cumulative infection in different groups [55]. Assays measuring functional neutralising ability might reflect protective immunity, based on animal experiments. Nonetheless, epitopes for HPV are not completely characterised in human responses therefore, they will be heterogeneous [55].  Herney Andrés García-Perdomo et al.
Neutralisation assays are independent of vaccine material, therefore they provide an unbiased measure of HPV serological response, induced by the vaccine. The direct ELISA, assesses the immunoglobulin G (IgG) response (both antibodies), that might be used as a technique to manage the large volume of samples [56].
On the other hand, the immune response to HPV infection (mainly IgG) is generally weak and heterogeneous among women. Around 50% of the individuals serologically convert to the L1 protein of HPV6, 16 and 18, within 18 months [57]. Conversely, it signifies that more than 40% of women do not seroconvert over time, therefore, the HPV L1 capsid-specific antibody is not an important diagnostic test for this infection. Other HPV antigens (E1, E2, E6 and L2) do not resemble any responses in patients with HPV infection [58]. But, HPV vaccines are based on recombinant virus-like particles (VLPs). These vaccines are highly immunogenic; induce very high titres of neutralising antibodies and represent durable responses. This might be due to

Epidemiology and Infection
VLPs since its ordered structure which permits the presentation of epitopes to B-cells for potent activation [59]. The challenge would be demonstrating that L1 gene expressed via a recombinant virus and the L1 protein produced largely and selfassembled into the VLP or the empty capsid almost identical to the native virion. These VLPs generate neutralising antibodies in animal models and these were protected against a virus challenge [60].

Strengths and limitations of the review
The advantages of this systematic review include: following standardised methods, according to Cochrane collaboration and PRISMA guidelines and the wide-ranging search to identify data about both clinical and immunological outcomes.
Although we had a lot of studies, most of our limitations were regarding available data for the vaccines across the studies. Additionally, we found differences in study design and methods to assess specific efficacy endpoints and immune responses [61].
On the other hand, we found high statistical heterogeneity and different issues regarding the risk of bias, mainly for not having enough information to evaluate the effect. However, some studies had a high risk of bias regarding blinding and attrition bias. All studies were industry-funded; therefore, this might lead to an unclear risk of bias.

Implications for policy and research
Hence, this is the first step to evaluate the efficacy of the vaccines against HPV. According to these findings, the vaccines reduce the risk of having the genotype that they were designed for. However, we need to go further, assessing the efficacy to reduce the risk of cancer. Although they were statistically valid, the risk reduction was not high (1-10%) compared with the general chance of not receiving intervention. Therefore, it is essential to consider this when decision-making in public health.
Another interesting data for public health is that there were similar adverse effects in both groups (vaccine vs. no intervention), therefore, even if risk reduction was not high, there were no increase of adverse effects in vaccinated people.

Conclusions
There was a reduction in the prevalence of HPV16, HPV18 and HPV16/18 genotypes when applying the bivalent vaccine, with no increase in adverse effects. Regarding the tetravalent vaccine, we found a reduction in the prevalence of HPV6, HPV11, HPV16 and HPV18 genotypes against placebo/no intervention, with no increase in adverse effects.
We suggest describing better and establishing measures to prevent attrition bias in RCTs regarding this critical issue. Financial support. None.
Ethical standards. This systematic review and meta-analysis accomplishes all the ethics requirements according to Helsinki declaration and all international statements.