The present study was not submitted to IRB because it was a systematic review exclusively with published articles.

Two authors performed searches of MEDLINE, The Cochrane Library, Embase, LILACS, and SciELO electronic databases. The search included published articles up to December of 2014 in Portuguese, Spanish, and English. A wide search strategy was employed in order to avoid publication biases and to assess studies dealing only with oral and/or oropharyngeal HPV infections in the Brazilian population. The following Medical Subject Headings and keywords were used: “human papillomavirus” and “Brazil” and “mouth” OR “human papillomavirus” and “Brazil” and “oropharynx”. Reference lists of previously obtained articles were also analyzed manually so that other relevant studies could be identified for inclusion in the present study.

Articles that were not published in Portuguese, Spanish, or English, or where the full text could not be obtained, were excluded from the search. Articles that had no data or insufficient data regarding HPV statuses were also excluded. Articles from the same institution by the same set of authors were screened for study time-period overlaps, and if repetitive information was presented, duplicated data were excluded and data from the most complete data set were analyzed. Other exclusion criteria were: articles that included only patients who tested positive for oral or oropharyngeal HPV; articles based on a pediatric population, animals, or cell lines; review articles, case reports, or case series studies; HPV in sites other than the mouth and/or oropharynx; articles that did not study HPV; and studies with populations other than Brazilians or multicentric studies in which it was not possible to separate the Brazilian population's HPV status.

All data were extracted by two independent authors using a data recording form developed for this purpose, which included the following: city and state, population characteristics, dates when the sample was obtained, the type of material collected (paraffin-embedded, fresh tissue, and/or brushing mucosa) and subsite (mouth and/or oropharynx), age, smoking status, alcohol intake, gender, technology used for HPV detection (polymerase chain reaction [PCR], in situ hybridization, and/or immunohistochemistry [IHC]), and general and specific HPV status (HPV 16 and/or HPV 18). Any discrepancies were addressed following discussion and consensus between the two authors.

The articles were initially organized in the following groups to facilitate the analysis: (1) oral or oropharyngeal HPV infection in healthy patients; (2) oral or oropharyngeal HPV infection in patients with benign or premalignant lesions; (3) HPV infections in the oral cavity and/or oropharyngeal squamous cell carcinoma (SCC); (4) oral or oropharyngeal HPV infection where patients or patients’ partners had genital HPV infections; (5) oral or oropharyngeal HPV infection in immunodeficient patients.

For the statistical analysis, the HPV status data were categorized into the following groups: healthy patients, immunodeficient patients, high-risk patients (those with genital HPV infections or infected partners), patients with benign lesions, with premalignant lesions (dysplasia, leukoplakia or erythroplakia), and with squamous cell carcinoma (oral cavity or oropharyngeal SCC). Microsoft Excel (Microsoft Corp® – Redmond, WA, United States) was used to tabulate data and to perform weighted average, frequency, and 95% confidence interval (95% CI) calculations. SPSS® version 17.0 (SPSS® Inc – Illinois, United States) was used for logistic regressions to assess the risk of HPV infection in each group, calculating odds ratios (OR) and 95% CIs. In all of these analyses, the probability of making an α or type I error was considered as less than 5% (p<0.05).

Eight articles (19%) studied oral or oropharyngeal HPV infection in healthy patients, two (4.8%) exclusively analyzed HPV status of patients with benign or premalignant oral lesions, 22 (52.4%) analyzed SCC patients (comparative or population studies), eight (19%) specifically studied oral or oropharyngeal HPV infections in patients with genital HPV infections or lesions or with infected partners, and two (4.8%) studied immunodeficient patients (Fanconi's anemia and HIV patients).

Thirty-four articles (80.9%) included demographic data, which are displayed in Table 2. It was observed that the weighted average for age was higher in the SCC group (59 years old) than the others (30–36 years old). Tobacco use and alcohol intake were similarly higher in these patients.

The 42 articles included 4066 enrolled patients. It was observed that oral or oropharyngeal HPV infections were identified in 738 patients (18.2%; 95% CI 17.6%–18.8%), varying between 0.0% and 91.9%. HPV status was assessed using PCR in 90.5% of the studies and by in situ hybridization in the others.

General HPV, HPV 16, HPV 18, and HPV 16/18 prevalences are displayed in Table 3. It was observed that higher rates of HPV infection occurred in patients with oral lesions and in people in high-risk groups (patients with genital lesions or infected partners). It was also noted that the healthy population had a very low HPV infection rate (6.2%; 95% CI 5.7–6.7%).

The risk of HPV infection was estimated for each subgroup (Table 4). It was evident that, when compared to the healthy population, the risk of HPV infection was approximately 1.5- to 9.0-fold higher, especially in patients with an immunodeficiency, oral lesions, and SCC. The rates of the most well-known oncogenic types (HPV 16 and/or 18) also show this increased risk.

The prevalence of oral HPV has been estimated to be in the range 2.4–7.5% among young adults in the United States.47 A recent study48 investigated the prevalence of oral HPV infections in the general US population (5579 men and women aged between 14 and 69 years) in 2009 and 2010. The overall prevalence of oral HPV infections for any HPV type was 6.9% (95% CI 5.7%–8.3%), the prevalence of high-risk HPV types was 3.7% (95% CI 3.0%–4.6%), and that of HPV 16 was 1.0% (95% CI 0.7%–1.3%). The present study identified a rate of 6.2% for any HPV infection and 1.4% for HPV 16 infection among 2060 healthy patients, which is similar to large series around the world.

Considering other groups analyzed in the present study, other studies also identified an increased prevalence of oral HPV infection. A study conducted in Zagreb, Croatia, established a rate of 17.7% for HPV positivity in oral lesions, with a significantly higher presence in benign proliferative mucosal lesions (18.6%). High-risk HPV types were predominantly found in potentially malignant oral disorders (HPV 16 in 4.3%) with a rate for any HPV infection of 15.5%.49 A meta-analysis conducted with dysplasia specimens from the oral cavity and oropharynx identified an overall prevalence of HPV 16/18 of 24.5% (95% CI 16.4%–36.7%).50 Another study identified a prevalence of 6.9% for oral HPV infection in HIV patients.51 Studies of this relationship in other immunodeficient patients are scarce. Oral HPV infections were found in almost 10.2% to 75% of patients with genital HPV lesions or infected partners worldwide.52,53

Genomic DNA of oncogenic HPV, especially type 16, has been detected in approximately 25% of HNSCC cases worldwide.54 In a large-scale meta-analysis, Kreimer et al. 55 reported an overall HPV prevalence of 25.9% in HNSCC (23.5% for oral cavity SCC and 35.6% for oropharyngeal SCC). In a recent French multi-institutional study, St Guily et al. 56 reported an overall prevalence of 46.5% for oropharyngeal SCC and 10.5% for oral cavity SCC. The present review established a similar rate of 27.4% for HPV infections in patients with SCC of the oral cavity or oropharynx and of 8.0% specifically for HPV 16 infection. For HPV 18 infection, a rate of 23.7% was observed among SCC patients in the present review, possibly the first report in this field with a large sample. Other series,55 however, reported an unexpected extreme rarity of HPV 18 in the oropharynx that was confirmed in other large studies, and the authors discussed how this was a difficult fact to explain.

O’Rorke et al.57 performed a meta-analysis with 42 articles studying the survival rate of HNSCC patients. They concluded that patients with HPV-positive HNSCC had a 54% better overall survival rate compared to HPV-negative patients (HR 0.46, 95% CI 0.37–0.57) and that both progression-free survival and disease-free survival were significantly improved in HPV-positive HNSCCs. This was not the aim of the present review, but it demonstrates the current relevance of the study of this topic.

Demographically, patients with HPV-related HNSCCs are more likely to be male, white, non-smokers, non-drinkers, younger in age, and have a higher socioeconomic status.57–60 Changes in sexual behavior, including a higher number of lifetime sex partners, an increase in oral sex practices, same sex-contact, and earlier age of sexual activity have all been implicated in HPV-related HNSCC.57,61 Many of the studies included in this review did not present findings by potentially important covariates (age, gender, tobacco use, alcohol use, etc.), as was observed in other studies.55,62 Moreover, only eight studies (38.1%) employed a cancer-free control group, which is difficult and can sometimes cause an overestimation of the odds ratio of HPV risk infection between HNSCC groups. The same trend was observed by Herrero et al.62

The variations in oral HPV prevalence among different studies may be due to differences in study populations, sampling and testing methods, and possibly the time periods studied.47

In the absence of a standardized detection technique, considerable variations in the prevalence of HPV infections have been recorded in the literature.54 This was not observed in the present review because the great majority of the studies employed GP 5+/6+ or MY09/11 primers in polymerase chain reaction (PCR) or HPV type-specific in situ hybridization.

The prevalence of HPV was assessed in a study with 1688 healthy men (median age of 31 years) in the U.S., Mexico, and Brazil. The prevalence of all HPV types was 4.0% (95% CI 3.1%–5.0%).10 The most prevalent high-risk HPV type was HPV 16 in all three countries; however, HPV55 was more common in Mexico compared to the U.S. or Brazil. The geographic heterogeneity might be partly explained by regional differences in the distribution of risk factors other than HPV infection in HNSCC patients.55

The study by Gillison et al.48 estimated an HPV infection rate of 6.9% in the general U.S. population in a sample obtained in 2009 and 2010. This rate was slightly higher than that observed in a systematic review that included data from 1997 to 2009 (4.5%; 95% CI 3.9–5.1).63

These results suggest that the prevalence of HPV may vary depending on distinct population characteristics and the time period studied, which might explain the wide range of HPV prevalence among different studies included in this systematic review.

There are a few limitations in this study that should be noted. The most important was in the data collection; because this work was a systematic review of published articles, some data could not be obtained, especially regarding demographics. Additionally, as seen in Fig. 2, the whole Brazilian population was not included, because of the lack of studies in some regions. Another important point is that because aggregate patient data were used, it is possible that there was heterogeneity in these studies and inconsistencies in the datasets that were unknown due to the summarization of the data. Moreover, the majority of the studies were small and used nonprobability samples, and it is difficult to differentiate studies that enrolled consecutive patients from studies that used alternative inclusion criteria, such as institutional archives of paraffin-embedded tissues. Nevertheless, these nonprobability samples did not represent the Brazilian population as a whole; however, the present study is the first one to consolidate the data of oral and oropharyngeal HPV infection in Brazil. HPV prevalence appeared to be inversely proportional to the study sample size, and the poor quality of some of the cancer specimens may also have affected the prevalence estimates. All of these points were also mentioned as limitations in other systematic review series.55