br Although EBV is the infection with
Although EBV is the infection with the highest prevalence with > 90% of adults infected (de-The et al., 1975), it 263717-53-9 was responsible for only 8.1% of the infection-caused cancers in Canada in 2015. In a similar vein, some infections with PARs of 100% were responsible for a small number of cancers (e.g. HHV-8 and HTLV-1) because of the rarity of cancers they cause.
4.3. Human papillomavirus
We found that 54% of the infection-associated cancers were due to HPV. This percentage is higher than the reported 29.5% global con-tribution of HPV to infection-associated cancers (Plummer et al., 2016). In particular our estimates for HPV16's role in head and neck cancers were higher than global estimates. Meta-analyses have reported higher HPV prevalence in oropharyngeal cancers in North American popula-tions compared to other continents (Ndiaye et al., 2014; Mehanna et al., 2013). Our estimate of 60.2% with E6/E7 detection, albeit numerically similar to that of Ndiaye et al. (60.4%) (Ndiaye et al., 2014), is actually higher than the latter because it represents detection of HPV16, whereas the 60.4% estimate in that study is for all HPV types combined. The oropharynx had the third highest number of attributable cases. Since 1997, oropharyngeal and oral cancer incidence has increased in Canada, especially among men, this is in part due to HPV's role in head and neck cancers (Canadian Cancer Society's Advisory Committee on
Cancer Statistics, 2015). The Canadian Cancer Society estimated that in 2012, cervical and oropharyngeal cancers each accounted for 35% of the HPV-associated cancer burden. We too, found that approximately one-third of the HPV associated cancer burden was due to cervical (35.9%) and oropharyngeal (28.3%) cancers. Since we examined the contribution of HPV16, any of the three available HPV vaccines provide coverage against this HPV type. Although a smaller proportion of oral cavity and laryngeal cancers are attributable to HPV16 (8.2% and 12.7%, respectively), they added 269 cases to the infection-associated cancer burden. School-based HPV immunization programs began in Canada in 2007. More recently, these programs have been extended to boys (Shapiro et al., 2017). We found that one-third (34.0%) of HPV associated cancers were diagnosed among men, which further empha-sizes the importance of vaccinating boys.
The main limitation of our study was the lack of Canadian-specific infection data and the subsequent reliance on data collected in the United States and for H. pylori data collected from European popula-tions. We have assumed that the exposure prevalence and strength of the relationship between the infection and cancer as observed in American and European populations were comparable to what would have been observed in Canada. For example, we reweighted the age, sex and race/ethnicity distribution from a population-based survey of H. pylori prevalence in the United States (NHANES) to match that of the Canadian population in the closest available year. Reweighting as-sumed that differences in the prevalence of H. pylori between the two countries were due to age, sex, and race/ethnicity – but these variables do not likely fully account for the potential differences between Canada and the United States. For some infection cancer site pairs, irrespective of including data collected outside of Canada and performing more targeted literature searches, the data remained sparse. This situation was particularly true for: H. pylori and gastric mucosa-associated lym-phoid tissue lymphoma, EBV and Burkitt lymphoma, and HPV and vaginal cancer. This result was anticipated since the cancer sites with sparser evidence were also the rarer cancers.
Focusing exclusively on Canada allowed us to obtain much of the rare and subsite cancer incidence data we required for accurate esti-mates of the number of attributable cases. However, we estimated ra-ther than directly obtained hepatocellular carcinoma and Quebec's cancer incidence. For cancer sites with fewer than 500 cases in Canada in 2015, the five-year incidence rates were averaged but this averaging relies on assumptions that the average of the last five years of available cancer incidence for Quebec (2006–2010) is representative of the 2015 cancer incidence, and that the trend has remained stable.
Since we used existing data, our findings inherited the methodologic flaws of included studies and population-based surveys. This concern was at least partially mitigated by including only those studies that met stringent inclusion criteria aimed at enhancing the validity of our es-timates. We attempted to correct for measurement error; however, some error may remain. Additionally, our correction for error in as-sessing the association between H. pylori and non-cardia gastric cancer did not account for confounders. Although the included studies were matched case-control studies, unmatched confounders have not been adjusted for.