The median age of the 9949 women for whom IgG antibodies to HPV VLPs were measured was 38 years (range: 18–97 years). Overall, seroprevalence for HPV-16, -18, -31, and -45 was 15, 15, 16, and 11%, respectively (
). Of the 1533 women seropositive for HPV-16, 579 (38%) were seropositive only for HPV-16 and 954 (62%) were seropositive for HPV-16 in addition to HPV-18 and -31. Similarly, for women seropositive for HPV-18, -31, and -45, the minority (41, 39, and 27%) were seropositive only for that type.
Seroprevalence of HPV-16, -18, -31, and -45 in the Costa Rican population-based cohort
As more than 60% of seropositive women were seropositive for multiple HPV types, we assessed type-specificity to distinguish between true multiple infections and potential crossreactivity of the assays. We assessed the association between seropositivity to one HPV type and that to another type (
). Seroreactivity to HPV-16, -18, -31, or -45 was strongly associated with seroreactivity to the other measured HPV types. Prevalence ORs appeared similarly elevated, although highest for HPV-18 and -45 (OR: 14.9; 95% CI: 11.4–19.4), HPV-31 and -45 (OR: 10.3; 95% CI: 8.0–13.3), and HPV-16 and -31 (OR: 8.5; 95% CI: 7.5–9.6). Analyses restricted to single infections defined by concurrent DNA resulted in diminished but still significant associations (data not shown).
Prevalence ORs and 95% CIs of HPV-16, -18, -31, and -45 seroreactivity vs seroreactivity to other type
To further clarify whether these associations reflect concomitant infections, we calculated the association between HPV seropositivity and viral DNA positivity for the same type and for the three other HPV types. As shown in
, for all four HPV types measured, the magnitude of the association was highest for each HPV serotype and DNA of the same type. The prevalence OR for HPV-16 seropositivity and HPV-16 DNA positivity was 4.5 (95% CI: 3.6–5.6). For HPV-18, -31, and -45 seropositivity and DNA positivity, the prevalence ORs were 2.7 (95% CI: 1.8–3.9), 5.1 (95% CI: 3.6–7.3), and 3.0 (95% CI: 1.2–7.2), respectively. The type-specificity of the serologic assay appeared weakest for HPV-18 and -45. When analyses were restricted to single infections of the specified HPV type by serology or by DNA, the magnitude of risk for serology and DNA of the same type remained similarly elevated, while the magnitude of association for serology and DNA of a different type declined to null, although confidence intervals were widened by reduced numbers. Moreover, when the analyses were restricted to HPV-exposed women, defined as either HPV DNA- or sero-positive for any of the four HPV types measured, thus reducing confounding due to (levels of relevant) sexual behaviour, the magnitude of association for HPV DNA and serology of the same type became much stronger. Again, CIs were widened due to reduced numbers. We also assessed the association between HPV-16, -18, -31, and -45 seropositivity with other HPV DNA types for which data were available (HPV-6, -11, -26, -33, -35, -39, -40, -51, -52, -53, -54, -55, -56, -58, -59, -61, -66, -68, -70, -71, -73); null associations largely were observed between HPV seropositivity with these other HPV types (data not shown), supporting type-specificity of the four serologic assays.
Prevalence ORs and 95% CIs of HPV-16, -18, -31, and -45 seroreactivity vs DNA-type positivity of the same type
In our population, HPV seroprevalence was significantly higher than HPV DNA prevalence, which was 3.5% for HPV-16, 1.3% for HPV-18, 1.4% for HPV-31, and 0.8% for HPV-45. As described in the Methods section, HPV DNA was not measured in study virgins; therefore, with the assumption that virgins are HPV DNA-negative (Kjaer et al, 2001
), the HPV DNA prevalence rates would be even lower. Of women HPV DNA-positive for each respective type, 45% were seropositive for HPV-16, 34% were seropositive for HPV-18, 51% were seropositive for HPV-31, and 28% were seropositive for HPV-45, thus reflecting the understanding that not all women infected with HPV will seroconvert, as defined within the detection limits of our assays and our stringent cutpoint. Conversely, of women seropositive for HPV-16, only 10% were HPV DNA-positive; for HPV-18, -31, and -45 seropositive women, DNA positivity for their same HPV type was 3, 4, and 2%, respectively, reflecting the transient nature of HPV infection.
HPV seroprevalence and DNA prevalence by age are shown in . At all ages, HPV seroprevalence of all four types remained elevated compared to DNA positivity. HPV seroprevalence reached its highest peak at 25–34 years for HPV-31 and at 35–44 years for HPV-16, -18, and -45. Although seropositivity appeared to decline slightly with age after its peak, seroprevalence always remained elevated above the level seen in women less than 25 years old. In contrast, HPV DNA positivity peaked for the four HPV types in women less than 25 years old and declined with increasing age; for HPV-16, however, there appeared to be a slight secondary increase in DNA prevalence in the older age groups. It is important to note again that these data for HPV DNA positivity do not include virgins, the majority of whom were less than 25 years (64%) or 25–34 years (20%).
Figure 1 Age distribution of HPV-16, -18, -31, and -45 seroprevalence* and DNA-prevalence** in Guanacaste, Costa Rica women. *Population-based HPV-16, -18, -31, and -45 serology prevalence is shown in black lines and includes study (more ...)
To identify determinants of HPV seropositivity, we assessed the association between demographic, social, and behavioural risk factors with HPV-16, -18, and -31 seropositivity in sexually active women (thus excluding virgins). Of all demographic and behaviuoral risk factors measured in univariate analyses, age, number of recent and past sexual partners, age at first intercourse, smoking, and OC use were statistically significantly associated with HPV seropositivity, and we therefore included them in our final multivariate model. As shown in
, there was a clear stepwise increase in prevalence and association with seropositivity for all four HPV types with increasing number of lifetime sexual partners. In our final model, the number of recent sexual partners, defined as within the past year, was no longer associated with HPV-16 seropositivity. Since having two or more recent partners was highly correlated with a high lifetime number of sexual partners, there was likely confounding by past exposure. Finally, for women who initiated sexual intercourse at older ages, statistically significant but small decreases in risk were observed for HPV-16, -18, and -31 seropositivity; this is likely due to the decreased lifetime number of sexual partners that is correlated with those initiating sexual intercourse at older ages.
Table 4 Final logistic regression model demonstrating association between number of lifetime sexual partners, age at first sexual intercourse, smoking, OC use and HPV-16, -18, and -31 seropositivity, also adjusted for age and recent sexual partners (excluding (more ...)
As shown in , current smoking exhibited a borderline association with HPV-16 seropositivity, with a risk of 1.2 (95% CI: 1.0–1.5); former and current OC use were associated with HPV-16 seropositivity with risks of 1.3 (95% CI: 1.1–1.5) and 1.5 (95% CI: 1.2–1.8), respectively. No association was demonstrated between HPV-18 seropositivity and smoking, but borderline associations were observed for current OC use, and between HPV-31 seropositivity and former OC use. Taken together, these results support the role of sexual behaviour as the key determinant of HPV seropositivity; the HPV cofactors for cervical cancer, namely OC use, may play a minor role in the risk of HPV-16 seropositivity. Determinants for combined HPV-16, -18, or -31 seropositivity similarly included number of lifetime sexual partners and age at first intercourse. However, neither smoking nor OC use was associated with combined seropositivity (data not shown).
We assessed the risk of CIN3/cancer associated with HPV exposure as determined by HPV DNA and serology measurements for HPV-16, -18, and -31. As shown in
, HPV-16 DNA-positive women possessed the greatest magnitude of risk for CIN3/cancer, regardless of serologic status, with age-adjusted ORs of 34.7 (19.7–61.0) for women both HPV-16 DNA- and sero-positive, and 39.9 (05% CI: 24.1–66.2) for women only HPV-16 DNA-positive. Women only seropositive for HPV-16 possessed a much lower, but still significantly elevated level of risk for CIN3/cancer, with OR of 2.0 (age-adjusted OR: 1.1–3.7). For HPV-18 and -31, the magnitude of association for women both DNA- and sero-positive with CIN3/cancer was the highest, followed by women DNA-positive only. Women only seropositive for HPV-18 possessed a moderate increase in risk for CIN3/cancer with an age-adjusted OR of 1.9 (95% CI: 1.2–3.0); women only seropositive for HPV-31 also possessed an increase in risk for CIN3/cancer (age-adjusted OR: 1.7; 95% CI: 1.1–2.7). Of CIN3/cancers seropositive for either HPV-16, -18, or -31, and DNA-negative for the respective type, the majority were also DNA-positive for another measured oncogenic HPV type at the time of the serological measurement, although it is unknown whether these are the same HPV types found in the tumour. When the analyses are restricted to single infections as defined by seropositivity, the magnitudes of association were slightly diminished for women seropositive but DNA-negative for HPV-31, but stayed constant for HPV-16 and -18 seropositivity.
Association between differential exposures (by serology and DNA) and disease outcome of CIN3/cancer, adjusted by age
We also conducted analyses stratified by age to delineate clearly risk relationships among young women, because past and recent exposures are likely equivalent in this age group. For women less than 25 years of age, the highest level of risk for CIN3/cancer was observed for HPV-16 DNA-positive women (OR=41.9) rather than DNA- and sero-positive women (OR=26.2). Higher risk estimates in DNA-positive women less than 25 years of age were also demonstrated for HPV-18. No statistically significant associations were observed between women HPV seropositivity only and CIN3/cancer for women less than 25 years of age. Exclusion of women less than 25 years of age did not significantly change the results.