Changes in HPV Seroprevalence from an Unvaccinated toward a Girls-Only Vaccinated Population in the Netherlands

Human papillomavirus (HPV), a virus capable of infecting the epithelial cells of the mucosa, is the cause of anogenital warts and cervical cancer (1). Besides cervical cancer, HPV is also linked to various other cancers in the anogenital tract and oral cavity (2). By routine HPV vaccination and effective cervical cancer screening programs, countries can reduce the burden of HPV-related disease.

All current globally available vaccines provide protection against HPV types 16 and 18, for example, and are included in the current bivalent vaccine. HPV types 6 and 11 are added in the quadrivalent vaccine and the nonavalent vaccine included additional HPV types 31, 33, 45, 52, and 58. Vaccination against HPV has been implemented in many countries, with the primary aim to protect women against cervical cancer. In the Netherlands, the bivalent HPV vaccine was implemented in the Dutch National Immunization Program as a girls-only vaccine for 12-year olds in a three-dose schedule in 2010, and is currently still being used, protecting them against HPV types 16 and 18. In addition, a catch-up campaign was initiated for girls from the birth cohorts 1993–1996 (i.e., 13–16 year olds) in 2009. From 2014 onward, the Netherlands shifted to a two-dose schedule (starting from birth cohort 2001 onward). The HPV vaccination coverage in girls in the Netherlands varied from 2009–2017 from 45% to 62% (3).

To gain information about previous HPV exposure, HPV serology is established as an important tool for population-based studies (4). This provides a view on type-specific cumulative lifetime exposure to HPV. Antibodies against HPV L1 virus-like particles (VLP) remain stable over time, and therefore reflect past infection and cumulative exposure. However, not everyone who contracted HPV will seroconvert, and the rate of seroconversion is known to be sex dependent (5). After HPV vaccination, HPV-specific antibodies are 10–100 times higher than (natural) infection-induced antibodies in serum (6), and therefore could be used to monitor vaccine uptake.

We assessed the (natural) infection-induced HPV seroprevalence for seven high-risk (hr) HPV types in the Dutch population in 2006/2007 (i.e., 4 years before the introduction of HPV) and 2016/2017 (i.e., 6 years postvaccination implementation). In addition, we investigated the effects of the introduction of the HPV vaccination on the seroprevalence of HPV types in our population.

Serum samples from two cross-sectional population-based serosurveillance studies performed from February 2006 to June 2007 and from September 2016 to October 2017 in the Netherlands were used for this study. Participants were 0–79 years of age in the 2006–2007 survey (n = 6,384), and 0–89 years of age for the 2016–17 cohort (n = 5,645). Study designs have been previously described in detail (7, 8). Briefly, the randomly invited participants were asked to fill in a questionnaire and to provide a blood sample. Questionnaires of both surveys included data on demographic characteristics, ethnicity (first- and second-generation migrants), vaccination history, and sexual behavior. Vaccination history was determined via the individuals' registration booklet and the Dutch vaccination registration Praeventis (9). The questionnaire used in 2006–7 was extended in the 2016–17 survey with more questions regarding sexual behavior. Information related to sexual behavior was only available from participants older than 14 years of age in both the 2006–07 study and the 2016–17 study.

We obtained written informed consent from all participants or their guardians before participation. The studies were conducted in accordance with recognized ethical guidelines (the Declaration of Helsinki) and were approved by an institutional review board “The Medical Ethics Committee Noord-Holland” in the Netherlands (METC number: ISRCTN 20164309 and M015–022).

Serum samples of both surveys were stored at −80°C until analysis, samples were measured at random for age and sex. For the measurement of HPV-specific IgG serum antibodies against L1 VLP of HPV16, 18, 31, 33, 45, 52, and 58, a VLP-based multiplex immunoassay was used as described previously (10). GSK (2006–07 survey) and MSD (Merck Sharp & Dohme; 2016–17 survey) produced the HPV-VLPs in used these studies. Briefly, VLPs were conjugated to seven distinct fluorescent microspheres via amine coupling. Serum samples were 1/50, 1/100, or 1/10,000 diluted and incubated with the VLP-coupled microspheres. HPV-specific IgG serum antibodies were detected using a secondary goat anti-human phycoerythrin-labeled antibody. Four in-house control sera and an in-house standard were used on each plate. The in-house standard (IVIG, lot LE12H227AF, Baxter) was calibrated against reference serum of GSK for all the seven HPV types. HPV-specific IgG antibodies were analyzed using the Bioplex system 200 with Bioplex software (Bio-Rad Laboratories). Samples were assumed to be seropositive above cutoffs according to the 99% Frey method (with 99% one-sided t values, based on concentrations measured in children of 1–10 years old (n = 859; ref. 11) and found to be 9, 13, 27, 11, 19, 14, and 31 Luminex units/mL (LU/mL; ref. 10) for HPV16, 18, 31, 33, 45, 52, and 58, respectively. As samples from 2016–17 were measured using a different batch of VLPs than those used in 2006–07, a correction formula was applied on the data of the 2016–17 survey. This correction formula was based on retesting of a random subset of 160 samples of the 2006–07 samples with the new VLPs. The correction formula was applied to the 2016–17 antibody measurements to align them with the 2006–07 measurements.

Data analyses were conducted using SAS version 9.4 and GraphPad Prism version 8.0.2. Women who were vaccinated against HPV according to the vaccination registry (n = 228) were excluded from analysis. In addition, women under 31 years of age and with arbitrary antibody concentration cutoff of >100 LU/mL for HPV16 and >50 LU/mL for HPV18 were considered to be “highly likely to have been vaccinated” and were excluded from the analyses (n = 18). Characteristics of the study population were compared among the 2006–2007 cohort and the 2016–2017 cohort using χ² tests. Seroprevalence for “any” or “all” hr-HPV-type(s) refer to the seven hr-serotypes that have been measured in this study. The study design (i.e., a two-stage cluster sampling method including specific regions and municipalities from which participants were invited) was taken into account in the analyses, as well as weights determined proportional to the reference population (Dutch population, January 1, 2007 and January 1, 2017, respectively) taking into account sex, age, ethnic origin, and urbanization degree. Seroprevalences were calculated per age-cohort and as large differences already have been observed between men and women (10), analyses were stratified for men and women. Crude seroprevalences of the different cohorts, age groups, and/or sexes were compared using Monte Carlo simulations. Parameters of the beta distribution for both seroprevalences were estimated and used in the simulations to obtain P values. Geometric mean concentrations (GMC) were calculated among HPV16 and HPV18 seropositive individuals from both cohorts, taking the study design into account. P values of <0.05 were considered statistically significant.

The associations between HPV seropositivity (positive for at least one out of the seven HPV types) in sexually active individuals older than 14 years of age who were not vaccinated and demographic characteristics (age, ethnic origin, degree of urbanization, education level, and socioeconomic status), was examined for the 2016–2017 cohort for men and women separately. In addition, associations with (sexual) behavior characteristics were taken into account, including: body mass index (BMI), alcohol consumption, smoking, having a steady partner, age of sexual debut (being defined as the first time of vaginal and/or penile intercourse), condom use at last sex act, number of partners in the last 6 months, lifetime number of partners, and reported history of STI (note: participants with missing values for a specific variable were allocated to a unknown category). We used generalized estimation equation (GEE) logistic regression models with a log link function and robust error variance. The incorporation of a GEE with exchangeable correlation structure accounted for dependency of multiple HPV types within an individual. First, univariate logistic regression analyses were conducted to study characteristics associated with HPV seropositivity. Variables that had P < 0.1 in univariate analyses were included in the multivariate analysis and backward selection (dropping variables one-by-one) was then applied. Hence, a multivariate model only including independently associated risk factors (P < 0.05) remained.

To study the differences in seroprevalence between the 2006–2007 and 2016–2017 cohort more closely, a pooled dataset was created including all HPV-unvaccinated participants from both cohorts. Again, the association between HPV seropositivity in sexually active individuals older than 14 years of age and demographics and sexual behavior characteristics was studied, in addition to the variable defining the cohort. Only characteristics available from both surveys were considered for inclusion in the model. Using a Poisson regression with robust error variance, we first calculated the crude prevalence ratio (PR). Next, we included the variables of interest to adjust for differences between the two surveys resulting in an adjusted PR (aPR). The analyses were performed for seroprevalence of any HPV type as well as type-specific. In addition, we stratified the analyses for men and women; we assumed that if herd effects on seroprevalence were to be observed this short after HPV vaccine introduction, this would be among men (first-order effect), in particular, younger males. Therefore, we looked also into the aPR for younger males (15–39 years of age).

We tested 5,645 serum samples, with corresponding response rates of 13.2% for men and 18.4% for women from the 2016–17 survey, and 6,384 serum samples, with corresponding response rates 28.9% for men and 34.7% for women from the 2006–07, which were tested previously (10, 12). Study characteristics were stratified for sex. In the 2016–17 survey, for both men and women, participants of 15 years and older were higher educated and had a higher net monthly income in comparison with both men and women in the 2006–07 survey. The mean age of sexual debut for people under 25 years of age was similar between the different surveys and sexes. However, age of sexual debut across all ages was lower in the 2016–17 survey compared with 2006–07 survey, for both men and women. In addition, the percentage of participants reporting to have a current steady partner was lower in 2016–2017, while “the number of sex partners in the last 6 months” and “ever having been diagnosed with a sexually transmitted disease (STD)” were higher in the 2016–17 survey compared with 2006–07 (Table 1).

An increase in seroprevalence for any hr-HPV type was observed in women in the age cohort from 15–19 years old, which reflects the median age of sexual debut. In the 2016–17 survey, seroprevalence for any type increased from 3.0% (10–14 years old) to 30.5% (20–24 years old) and 33.7% (25–29 years old) and peaked at 37.0% in the 30–39 year old. The greatest rise was seen for HPV16 and HPV18. This increase in seroprevalence was much more gradual for men, and mainly in the 2006–07 survey was most pronounced for any hr-HPV type and HPV16 (Fig. 1). Samples sizes of age cohorts can be found in Table 1 and Supplementary Table S1.

Low seroprevalences were observed in children 0–14 years of age in both sexes and surveys. In the 2016–17 survey, the highest seroprevalences in children (0–14) were detected for HPV16 and HPV18 (Fig. 1).

In the older female age groups, overall seroprevalence decreased from the age of 49 years onward in the 2006–07 survey and from 60–69 years onward in the 2016–17 survey. Age-specific higher seroprevalence for any hr-HPV type was observed in 2016–17 compared with 2006–07, being significant in age groups 30–39 (P = 0.0108), 50–59 (P = 0.0406), and 60–69 (P = 0.0056) years of age. A lower seroprevalence for any hr-HPV type was observed in the age group 10–14 (P = 0.0118) in 2016–17 compared with 2006–07. In men, a lower age-specific seroprevalence for any hr-HPV type was observed in 2016–17 compared with 2006–07, being only significant in the age groups 10–14 (P = 0.0056). No significant difference was found for the age-specific seroprevalence for any hr-HPV type excluding 16 and 18 in the age groups 10–14 and 15–19 years of age between the two surveys, P = 0.091 and P = 0.1206, respectively.

Unvaccinated female participants older than 15 years of age showed significantly higher seroprevalence for any hr-HPV type in 2016–17 compared with 2006–07; 31.4% [95% confidence interval (CI), 29.1–33.7] and 25.2% (95% CI, 23.1–2.3), respectively. For men from 15 years of age and older, seroprevalence for any hr-HPV type was similar between the 2006–07 and 2016–17 surveys; 19.7% (95% CI, 17.9–21.6) and 20.3% (95% CI, 18.4–22.1), respectively. In women, also seropositivity for one up to all seven types was significantly higher in 2016–17 and hr-type specific. Type-specific HPV16, HPV18, HPV31, and HPV58 were higher in 2016–17 compared with 2006–07, which was also true for the combinations HPV16 and 18, HPV16 or 18, and HPV16 and/or 18 (Table 2).

For men, this was true for the combination HPV16 or 18, positivity for more than two hr-HPV types and type-specific HPV18, 31, 33, 45, 52, and 58. For HPV16 a lower seroprevalence was seen in 2016–17 (7.5%; 95% CI, 6.5–8.5) compared with 2006–07 (10.6%; 95% CI, 9.2–12.0). Just as for the combination HPV16 and 18, HPV16 and/or HPV18 and positivity for more than one hr-HPV type (Table 2). HPV16 was also most prevalent in both surveys, followed by HPV18, HPV45, and the rest of the types (Table 2). Only a very small percentage of the males were seropositive for all seven hr-HPV types, 0.6% and 0.3% for 2006–07 and 2016–17, respectively.

The age-specific HPV16 GMCs of (natural) infection-induced seropositive women as well as of seropositive men were comparable in all age cohorts between both studies. No differences were found between the GMCs of the HPV16 and HPV18 seropositive individuals between 2006–07 and 2016–17 (Supplementary Fig. S3).

For women, the univariate analysis showed an association for HPV seropositivity for any hr-HPV type with middle educational level, being a first- or second-generation migrant, having a lower income, ever used alcohol, not having a steady partner, lower age of sexual debut, having more than two sexual partners last 6 months, history of reported STD, and having more than two sexual partners during lifetime. In the backward selection model, low and middle educational level, first-generation migrants, more than two sexual partners during lifetime, and history of self-reported STDs remained and were independently associated with seropositivity for any hr-HPV type (Table 3A and B).

For men, the univariate analyses showed only an association of seropositivity for any hr-HPV type with history of self-reported STDs (Table 3A and B).

HPV seropositivity for any hr-HPV type for women from 15 years onward was 25.2% in 2006–07 and 31.4% in 2016–17. After pooling both surveys, and adjusting for demographic characteristics (age, sex, urbanization, education, income, ethnicity) and sexual risk factors (age of sexual debut, number of partners during the last 6 months, history of STDs), this resulted in a smaller, but still significant, difference between 2006–07 and 2016–17, (aPR 1.16; 95% CI, 1.02–1.32). Before adjustment of any variables, all seven hr-HPV types were significantly higher in 2016–17 compared with 2006–07. However, after adjustment for the demographic characteristics and sexual risk factors, the differences remained only significant for HPV16 (aPR 1.29; 95% CI, 1.07–1.55), HPV18 (aPR 1.31; 95% CI, 1.01–1.70), HPV31 (aPR 1.54; 95% CI, 1.111–2.14), and HPV52 (aPR 1.27; 95% CI, 1.00–1.62; Table 4).

For men from 15 years onward, HPV seropositivity for any hr-HPV type was significantly lower in 2016–17(18.2%) compared with 2006–07 (20.3%). After adjustment for the demographic characteristics and sexual risk factors, this did not remain significant (aPR 0.99; 95% CI, 0.83–1.17). Before adjustment, HPV16, HPV18, and HPV31 were significantly lower in 2016–17 compared with 2006–07 and HPV45 was significantly higher in 2016–17 compared with 2006–07. This difference only remained significant for HPV16 after adjustment (aPR 0.71; 95% CI, 0.55–0.91; Table 4) and HPV45 (aPR 1.47; 95% CI, 1.13–1.92). Zooming in on men in the age cohort of 15–39, a nonsignificant decrease for HPV16 between 2006–07 and 2016–17 was observed (aPR 0.84; 95% CI, 0.52–1.37).

In this study, we assessed the (natural) infection-induced seroprevalence of seven hr-HPV types in the Dutch population before and 6 to 7 years after the introduction of a girls-only bivalent HPV vaccination program, with an uptake varying over the years between 42% and 61%. Surprisingly, HPV seroprevalence in female age cohorts of 15 years and older has increased in a 10-year time period, mainly due to a significant increase in HPV16, 18, 31, and 58. In men, however, seroprevalence for any hr-HPV type remained similar with a decreasing trend found for HPV16 and increasing trend for HPV45.

We restricted the analyses to unvaccinated individuals thereby estimating naturally acquired and cumulative type-specific HPV exposure. The increase in seroprevalence with age for women was in line with the age of sexual debut. The peak in HPV seropositivity was highest in women ages 30–39, which has been reported in other publications (13–15). This peak in seroprevalence around 10 to 20 years after sexual debut might reflect repeated exposures resulting in a subsequent increase of the seroconversion rate to induce a detectable antibody response (16).

In the 2006–07 survey, the seroprevalence in middle-aged and older women declined at an earlier age than in the 2016–17 survey, where levels started to decrease from 70 years and onward. In males, this decline in HPV seropositivity is not seen. The slight decrease of seropositivity observed in older women could be explained by waning of antibodies which was suggested by af Geijersstam and colleagues (4). This would mean that seroprevalence is underestimating lifetime cumulative exposure. Alternatively, it could reflect a cohort effect, which is more likely as this effect is not seen in both sexes. Indeed, although age of sexual debut is similar, the sexual behavior pattern differs in the younger women having more lifetime sexual partners than the older women in this cohort (17).

Hr-HPV antibodies, albeit at very low concentrations, could be detected in children, which confirm other population studies (10, 16). These antibodies might be derived from vertical or horizontal transmission (18).

In both surveys, a lower seroprevalence is observed in males compared with women, which conform other population studies (10, 13, 15, 16, 19). It is unlikely that the overall lower seroprevalence seen in males is due to lower infection rates, because males reported a significantly higher number of lifetime sexual partners compared with females in the 2016–17 cohort. HPV DNA prevalence studies showed similar results among both sexes (20, 21). The fact that women display a higher seroprevalence then men is likely to be explained by the anatomic site of the HPV infection, influencing its immune response. Infections at the epithelium of the cervix and anal tract induce higher immune response in comparison with infections that occur at the keratinized epithelia, such as genital skin (22–25).

Seroprevalence for HPV16 was highest of all HPV types in both surveys, which is in accordance with other population studies (15, 16, 26–29). In a 10-year time period, HPV type-specific seroprevalence for HPV16, 18, 31, and 58 has increased in the female population of 15 years and older in the Netherlands. In addition, being seropositive for one up to all seven types increased over the years. This is possibly explained by the observed change toward a higher number of sexual partners in the last 6 months and history of self-reported STDs in the 2016–17 survey compared with the 2006–07 survey.

The risk factor analysis was restricted to the HPV unvaccinated, sexually active population from 15 years onward and stratified for sex. For women several behavioral factors, such as number of lifetime partners, history of STDs, and ethnicity, were independently associated with HPV seropositivity, which was also found in other studies (10, 15, 30, 31). Alcohol use and smoking were only associated with HPV seropositivity in the univariate model. This is especially interesting with respect to the current increased risk in HPV-associated head and neck cancers (32, 33). In these studies, alcohol use is often allied together with tobacco use; however, in our study, we could not find an association with smoking or alcohol use in the multivariate model. Other studies find varying results, showing a negative association (29), a positive association (14, 34), or no association at all (35). Thereby leaving the relation between smoking and HPV seropositivity unclear.

In the male part of our study, we only found history of self-reported STDs to be significantly associated with HPV seropositivity in the univariate analyses. In the backward selection model, this factor did not remain independently associated with HPV seropositivity anymore. Comparison with other studies is challenging as most population studies combine men and women in their risk factor analysis. Studies including separate male analysis reported a variety of associated factors. Most consistent findings were associations related to age (13, 31, 36–42), number of male anal sexual partners (30, 40, 41, 43), and (self-reported) circumcision (44). Number of male anal sexual partners and (self-reported) circumcision were unfortunately not included in our questionnaire. In addition, some of these studies were performed among men who have sex with men, which is considered a high-risk population with specific behavior, complicating direct comparison.

Also after pooling both surveys and adjusting for demographic and sexual risk factors, the increase in HPV seroprevalence in women in the years after the introduction of the HPV vaccine remained significant.

An interesting finding is the decreased HPV16 seroprevalence between the two surveys for the Dutch male population. Although this might be due to herd immunity of the girls-only vaccination program, among 15–39 year old men, we observed a not statistically significant decline, while they seem likely to be the first age groups benefitting from the girls-only vaccination. In the even younger age group of 10–14 year old boys, we did find a difference between the two surveys for any hr-HPV type, but when excluding the vaccine types this difference was not significant anymore. This indicates that the. Albeit minor, difference in seroprevalence in the 10-year period was mainly attributable to the vaccine types. Please note that in this age group, the seroprevalences could not be adjusted as questionnaires including sexual behavior were only filled in by people above 15 years of age. In Australia, with high vaccine uptake percentages, herd protection impacts on seroprevalence in males (15–39 years of age) from a girls-only vaccination program were clearly shown 5 years after introduction of HPV vaccination (45). Moreover, even a benefit for the nonvaccinated females was observed (46). In contrast, no reduction was found in HPV seropositivity in males followed by a girls-only vaccination program in a study in the United States with comparable vaccination coverage in a girls-only program as in the Netherlands (47). With a vaccination coverage of approximately 50% among vaccine-eligible girls in the Netherlands in 2016–17, herd effects on seroprevalence in the male part of the population might therefore be less pronounced. Nevertheless, in recent analysis among STI clinic visitors in the Netherlands, both first-order herd immunity effects among unvaccinated males as well as second-order herd immunity effects in unvaccinated women were found (48). However, this was measured through infection rates where effects can be detected earlier than by seroprevalence.

A strength of this study is that we compared two surveys with a broad age range, one before and one 6 to 7 years after introduction of the HPV vaccination program, thereby enabling us to evaluate this program at a population level. An additional strength of this study is that an identical methodology and antibody assay is used between the two surveys. It must be kept in mind that the VLP sources have changed over time and could possibly cause some variance, for which we corrected by using QC and bridging. Furthermore, this study used two-stage cluster sampling strategy, including oversampling of minorities (7, 8), therefore being representative of the total Dutch population.

The use of different techniques and associated cutoff levels hinders the comparison with other (population) studies. International standardization for all hr-HPV types, in addition to HPV16 and 18 which are applied in this study, could help to overcome this difficulty in future studies. Besides this, it could be argued that HPV seropositivity is not a conclusive marker for cumulative exposure. HPV has the capability to evade the host immune system and as a consequence, detectable HPV-specific antibodies in serum are only developed in approximately 50%–70% of HPV-infected individuals (22). Thus, seroprevalence will underestimate the actual lifetime HPV exposure and infection rate. Moreover, it should be noted that questions regarding sexual behavior were among the least well completed. Self-reporting of sexual behavior could lead to bias due to social desirability and this was also illustrated by our risk factor analysis for some variables (e.g., a high unknown category).

HPV prophylactic vaccination programs are most effective when offered to nonsexually active preadolescents. In the Netherlands, the age of receiving the HPV vaccine is approximately 12 years old and recently the Health Council has advised to lower the age to 9 years. Our data support this change as HPV seropositivity begins to increase markedly after 10 years of age. In addition, the Health Council also advised to implement a sex-neutral HPV program which will be effective from 2021. On top of that, a catch-up vaccination will be offered to all young adults up to 26 years of age (49).

To conclude, our data showed that HPV infection-specific seroprevalence in women has increased in the Netherlands in a 10-year period. In men, however, seroprevalence for any hr-HPV type remained similar, a decrease was found for HPV16 and an increase for HPV45. Whether the decline in HPV16 is a first sign of a herd effect remains uncertain because a less pronounced effect was observed in men ages 15–39 years of age, where we would have expected that a herd effect would be visible first. Future seroprevalence studies will be interesting to capture the effect of a longer follow-up period after introduction of the girls-only program and possibly effects of the sex-neutral vaccination.

No potential conflicts of interest were disclosed.

H. Pasmans: Conceptualization, formal analysis, investigation, methodology, writing–original draft. J. Hoes: Formal analysis, writing–review and editing. L. Tymchenko: Data curation, investigation. H.E. de Melker: Conceptualization, supervision, writing–review and editing. F.R.M. van der Klis: Conceptualization, supervision, funding acquisition, project administration, writing–review and editing.

The authors thank GSK (GlaxoSmithKline Bio) for providing at cost part of the VLPs and MSD (Merck Sharp & Dohme) for kindly donating part of the VLPs that we used in our assays. GSK and MSD were provided the opportunity to review a preliminary version of this manuscript in order to ensure the protection of its proprietary information and intellectual property, but the authors are solely responsible for final content and interpretation. The authors also thank Eric Vos, Mirjam Knol, and Liesbeth Mollema for their help with the statistical analyses.

This research was supported by the Dutch Ministry of Health, Welfare and Sport. This study was supported by the Dutch Ministry of Health, Welfare and Sport (VWS). The funding source had no involvement in study design, collection, analysis, and interpretation of data; in writing of the manuscript; and in decision to submit the article for publication.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.