Trends in HPV cervical and seroprevalence and associations between oral and genital infection and serum antibodies in NHANES 2003–2012

The relatively low oral–cervicogenital type–concordance overall, noted by Steinau (2014) [14] and others, suggests that, although not independent, the sites have differences in natural history (e.g. time to clearance). Here we show that oral prevalence and oral–cervicogenital type-concordance vary dramatically with age. This concordance peaks for women ages 18–24, for whom nearly all oral infections are type-concordant (Fig. 1). This may suggest that many 18–24 year olds are experiencing their first oral infection and that it is either caused by autoinoculation from their genital infection or that both infections are from the same sexual partner. That we do not see the same pattern among ages 14–17 may be indicative of different sexual norms and practices between the two groups, a possibility that warrants further study. There is another age span, 45–54, with both high overall prevalence and high concordance. This may be indicative of a second sexual debut or may reflect cohort differences.

Although discussion of HPV in the public health context is often limited to mucosal types, other strains, including those that cause common and plantar warts, infect cutaneous tissue. It is possible that this tropism, that is preference for certain kinds of tissue, may be relevant within the mucosal types, with certain strains preferring genital tissue and others preferring oral tissue. It is difficult to determine whether some strains—such as types 16, 62, 84—are commonly found at both sites because of a lack of tropism, or simply because they are relatively prevalent. However, the pattern of prevalence for type 44 is consistent with a tropic preference for oral tissue. More work will be needed to assess whether this and other patterns—such as the higher prevalence of types 53, 54, and 61 among those with cervicogenital infections than among those with oral infections—can be attributed to tropism or are a reflection of cohort/timing patterns. Tropism may be important when considering an evolutionary perspective; changing sexual behaviors [36–39] may have created evolutionary pressure, assuming some level of genotype competition, favoring less tropic species.

Among those women who are seropositive for 6, 11, 16 or 18, prevalence of type-concordant cervicogenital infections is low and decreases with age, as might be expected since older individuals are more likely to have already seroconverted during previous infections. For racial demographics, seroprevalence more closely follows the cerivcogenital rather than oral profile. This finding may simply be a result of cervicogenital infections being much more common than oral ones for women, but it also raises the possibility that seroconversion may occur more readily following cervical rather than oral infection. Combined with the different prevalence of serotype-concordance for men and women (Fig. 4; the small racial deviations in the overall patterns can be attributed to low numbers and incomplete serotyping), this observation lends evidence to the theory that seroconversion is strongly site-specific and occurs primarily at mucosal epithelia. Cervicogenital infections appear to often lead to seroconversion, which, based on the lack of sero–oral type-concordance in women, may provide a defense against oral infection. Men are less likely to seroconvert from genital infections [17, 40–42], and so may be more vulnerable to oral infections, resulting in a higher oral prevalence.

Seroprevalence increases dramatically for 14–17 and 18–24 year old women over the period studied. Since the aim of the vaccine is seroconversion, it is likely that this increase is largely caused by the introduction of the vaccine (Fig. 5). Further, we want to determine the relative risk of infection given vaccination. If one does not control for age, overall cervicogenital prevalence and prevalence of oncogenic types is higher in the vaccinated population than in the unvaccinated, and prevalence of the genotypes targeted by the vaccine is only slightly lower among the vaccinated group. Because the vaccine is relatively new, it may be being given to previously infected individuals. If we restrict our attention to either 14–17 or 18–24 year olds, as in Table 2, the relative risk for any and oncogenic HPV drops to near 1, and that of the vaccine types drops to about 0.2. Our results follow those reported by Markowitz (2013) [43], who found that, in NHANES 2007–10, prevalence of vaccine types was 12.6 % among unvaccinated 14–19 year olds but 3.1 % among those vaccinated. These results suggests that the vaccine has been highly effective at preventing infections by the targeted types. The current data neither support nor refute the possibility that other genotypes may move in to fill the niche left by the vaccine genotypes.

Non-Hispanic black men and women have consistently high prevalence at genital and oral sites and seroprevalence, which is consistent with previous work [14, 32]. The APC cohort trends for cervicogenital prevalence (Fig. 6) suggest that this high prevalence relative to the other racial groups has not changed much between the 1955 and 1980 birth cohorts and has, in fact, been increasing since the 1980 birth cohort. Sexual assortativity patterns and other contributing factors may have been relatively consistent between the 1970s and 1990s, the times that people from these birth cohorts would have been mostly sexually active. The relative prevalence among women of all demographics decreased after the 1940s birth cohorts, the cohorts that experienced the sexual revolution of the 1960s. For women of all races, cohort effects for seroprevalence for women are dominated by a large increase for those born after 1980, most likely a consequence of vaccine-induced seroconversion. That age–cohort models fit better than age–period is not surprising, both as trends in STI prevalence tend to be driven by cohort sexual norms and as the NHANES time span is limited.

One strength of this study is the large sample size in the NHANES survey, which allowed analysis at a relatively fine demographic stratification. NHANES is the only population data source for oral infections and also allows for analysis of multiple sites in one individual. Additionally, the use of APC models allows for the analysis of temporal trends in the data separate from the age effects. Limitations of the NHANES data set include the relatively limited time span, especially for oral infections, the limited number of serotypes, and the lack of genital prevalence for men and anal prevalence for men and women. Although, to our knowledge, no studies have investigated seroconversion in women following anal infections, anal infections may play an important role in seroconversion in men [16, 17, 42].