There are currently 4 potentially vaccine-preventable human papillomavirus (HPV) types with 2 available vaccines—1 bivalent, the other quadrivalent. HPV types 16 and 18 are associated with approximately 70% of cervical cancer cases worldwide,1 25% to 80% of other anogenital cancers (e.g., anal, penile, and vaginal), and some head and neck cancers (e.g., mouth and oropharynx).2 HPV types 6 and 11 are associated with a significant burden of disease due to anogenital warts and the rarer but more severe condition of recurrent respiratory papillomatosis.3 The English national immunization programme was introduced in 2008, using the bivalent Cevarix vaccine providing protection against HPV types 16 and 18 for girls aged 12 years, with catch-up for girls aged up to 17.4
HPV DNA testing of cervical specimens gives a highly sensitive measure of current infections,5–7 most of which are transient and are cleared within 2 years.8 Amongst the approximate 60% of infected females who seroconvert,9 and lower percentage of males who seroconvert, antibodies offer a longer-term marker of infection. Seroprevalence studies can therefore provide estimates of cumulative exposure to HPV infections and have been used to describe the epidemiology of HPV in other countries10–12 and in young females in England.13 Interpretation of seroprevalence studies is, however, complicated by the lack of robust evidence about seroconversion and persistence of antibody responses, and the extent to which age-cohort dependent changes in risk of infection may affect cross- sectional studies. Antibody titers can help understand how seroprevalence, and possibly immunity, wanes after infection. Persistence of antibody responses following infection is known to vary by HPV type, with significantly more persistent responses for 16 and 18 than for HPV 6.9
We have examined type-specific HPV seroprevalence and antibody titers before vaccination amongst a population-based sample of males and females aged 10 to 49 years (including females aged 10–29 years as described previously13), and used mathematical modeling to investigate factors which could explain seroprevalence trends. This study offers an opportunity to improve our understanding of HPV epidemiology before it is altered by vaccination and to provide a baseline against which to evaluate the impact of the national HPV immunization programme.
MATERIALS AND METHODS
Serum specimens were obtained from the Health Protection Agency (HPA) Sero-epidemiology Unit (SEU) as described previously.14 Briefly, unlinked residual sera from routine microbiologic and biochemical investigations were submitted to the SEU by participating laboratories in England. Sera collected in 2002–2004 from males and females aged 10 to 49 years were selected from the SEU by gender and single year of age from 12 laboratories across England. The younger age groups (10–19 years) were oversampled to ensure about 180 samples for each single year of age. About 120 samples were selected for each year of age between 20 and 39 years, and 60 samples for each year of age between 40 and 49 years. The sample size was determined to provide an estimate of seroprevelance in females within 10% of the true seroprevalence (i.e., between 11% and 15% for expected seroprevalence of 13% for HPV 16), and to show age-specific trends by gender. The rationale for oversampling younger ages was the lower expected prevalence, the absence of other prevalence data, and the relative importance of data for these younger ages to inform immunization policy. Older ages (more than 40 years) were included in order to give a complete age-specific seroprevalence curve, rather than to give precise estimates for these age groups.
Serum samples were tested for antibodies to HPV types 6, 11, 16, and 18, using a multiplexed competitive Luminex assay as previously described.15 Testing was conducted at Merck & Co. Inc. Laboratories (females, 10–29 years) and at the HPA Vaccine Evaluation Unit in Manchester (all other samples) using critical reagents supplied by Merck. Quality assurance samples from Merck were run alongside the testing at HPA, with no differences demonstrated between the assay as run by Merck or by the HPA. Briefly, HPV specific virus-like particles (VLPs) were coupled to fluorescent Luminex beads. Sera were incubated with the beads and with HPV type-specific monoclonal antibodies, which compete with the serum antibodies for binding to sites on the VLPs. The assay measures the ability of the type-specific serum antibodies to prevent binding of the monoclonal antibodies to the VLPs. Antibody levels were expressed as milli-Merck units per milliliter (mMU mL−1). Sera were assumed to be seropositive at cutoffs determined in previous work with this assay16: 20, 16, 20, and 24 mMU mL−1 for HPV 6, 11, 16, and 18, respectively.
All analysis was conducted separately for males and females. Seroprevalence with 95% confidence intervals (CI) was calculated by single year of age and 5-year age groups for: (i) type-specific, (ii) any HPV type (6, 11, 16, or 18), (iii) high-risk (HR) (16 and/or 18), and (iv) low-risk (LR) (6 and/or 11) infections. Seroprevalence estimates were weighted to reflect the single-year age distribution of 10- to 49-year old males and females in England in 2004 using population figures from the Office of National Statistics.
Associations between type-specific HPV seropositivity and sample characteristics including 5-year age groups, year of sample, and region (north/south) were evaluated by univariable and multivariable logistic regression analyses. Pearson χ2 tests were used to test for differences in seroprevalence between subgroups. Likelihood ratio tests were used to identify variables significantly associated with seropositivity. Adjusted odds ratios and 95% CIs were reported for variables that were significantly (P < 0.05) associated with seropositivity.
To examine the distribution of antibody titers in seropositive individuals, the geometric mean titers (GMT) with 95% CI were calculated for each HPV type for 5-year age groups. Linear regression of log-transformed antibody levels was used to test whether the ratio of 2 GMTs (e.g., by age groups) was significantly different from one.
Statistical analyses were conducted using STATA 11.0 (StataCorp, College Station, TX).
A range of catalytic models were developed to investigate the potential cause of the observed sex-specific trends in seroprevalence with age. Catalytic models, which represent the process of moving from susceptible to seropositive states, are commonly used to estimate force of infection (FOI), using seroprevalence as a marker of past infection.17 Such models assume that a FOI acts upon an age-stratified population that is susceptible at birth. As individuals become infected, they are removed from the susceptible class. These models were adapted to investigate whether the following factors explain seroprevalence trends: (i) age-varying FOI (using an exponentially damped quadratic function similar to that used in previous HPV modelling work18), (ii) waning of antibody titers over time, and (iii) different risks of infection by age cohort due to changes in sexual behavior through time.
As the samples were unlinked to any behavioral information, changes in sexual behavior over time were estimated from the second National Survey of Sexual Attitudes and Lifestyles (NATSAL II) conducted in 2000–2001.19 The median age of first sex reported by 21 to 44 year olds in 5-year age groups was used as a proxy for level of sexual behavior and risk of HPV infection in each birth cohort. We only included data on individuals with a reported age of first sex of less than 20 years for all age groups, thus removing individuals with an age of first sex older than the youngest age group to allow complete comparability. The median age of first sex for age groups less than 20 years and more than 45 years was generated by either linearly extrapolating the trend in 20 to 45 year-olds, or assuming it had stabilized. The effect of the other factors that may explain seroprevalence trends were estimated through fitting. A separate model for each HPV type and gender was generated. Each model was fitted to seroprevalence data using maximum likelihood estimation. The models were then compared using the Akaike Information Criterion (see Text, Supplemental Digital Content 1, online only, available at: https://links.lww.com/OLQ/A12, for full methods).20
Sera from 4647 individuals were tested for HPV infection. More than half (65.6%) were collected during 2004 (15.6% from 2002, 18.9% from 2003) and 54% were from laboratories in north England (Leeds, Liverpool, Preston, and Manchester). Sera from south England were mostly from Dorchester, London, and Cambridge. Fifty-eight percent of samples from the south and 43% from the north were from males.
Seroprevalence for HPV 6, 11, 16, and 18 was 16.4%, 5.7%, 14.7%, and 6.3%, respectively, among females and 7.6%, 2.2%, 5.0%, and 2.0%, respectively, among males. Seroprevalence for any 1 type (6, 11, 16, or 18), HR, and LR types was 29.3%, 18.1%, and 19.1%, respectively, for females and 13.5%, 6.6%, and 8.5%, respectively, for males. Seroprevalence was significantly higher among females than males for each HPV type (P < 0.001 for each of the 4 HPV types).
Amongst females, seroprevalence by single year of age for every HPV type was very low until 15 years. After this age, seroprevalence of HPV 6 and 11 increased until 24 years, whereas HPV 16 and 18 seroprevalence continued to increase until the late 20 s and early 30 s after which it reduced in older ages (Fig. 1A). Seroprevalence for HPV 6 was significantly higher than HPV 11, and HPV 16 was significantly higher than 18 (P < 0.001 for both comparisons). In males, type-specific seroprevalence remained below 3% until age 19 and then rose until about 20 to 24 years of age after which it stabilized. It was lower for males aged 15 to 29 years than for females of the same ages (P < 0.02 for each of the 4 HPV types). In contrast to females, there was no fall in seroprevalence in older males. HR and LR seroprevalence was similar among females until the older ages where seroprevalence was lower for HR types than LR types (Fig. 1B). Among males, seroprevalence for both remained stable between ages 20 and 40 years, after which there may be a slight increase.
Seroprevalence was significantly higher in the south of England for HPV 16 among females (16.4% vs. 12.8%) and for all types except HPV 11 among males (8.6% vs. 5.5% for HPV 6, 5.7% vs. 3.5% for HPV 16, and 15.7% vs. 9.1% for HPV 18). There was no evidence of changes in seroprevalence between 2002 and 2004; we did not analyze date of collection further and considered this as a single cross-sectional sample. Seropositivity for each HPV type was associated with seropositivity for all other HPV types (P < 0.003 for all comparisons), regardless of gender.
In multivariable analysis, seropositivity to at least 1 type remained associated with being seropositive with another type for females and males after adjusting for age group (Table 1). The strongest association between HPV types was between HPV 11 and 6, where the odds of HPV 6 seropositivity were almost 5-fold higher in HPV 11 seropositive females and 12-fold higher in HPV 11 seropositive males (P < 0.001 for both). Similar associations were observed vice versa (data not shown). Increasing age remained significantly associated with seropositivity, for all HPV types and both genders (P < 0.001).
Distribution of Type-Specific Antibody Titer Among Seropositive Individuals
The overall GMTs (95% CI) for HPV 6, 11, 16, and 18 were 58.6 mMU mL−1 (53.9–63.6), 37.3 mMU mL−1 (31.9–43.6), 96.6 mMU mL−1 (86.7–107.7), and 69.3 mMU mL−1 (59.6–80.5), respectively, among females, and 49.2 mMU mL−1 (43.6–55.6), 56.8 mMU mL−1 (42.9–75.2), 89.9 mMU mL−1 (71.5–113.0), and 61.8 mMU mL−1 (45.8–83.3), respectively, among males. There was no significant difference in titers between males and females for any HPV type. There was no evidence of declining antibody titers with increasing age, for any of the 4 HPV types in either males or females (Fig. 2).
Titers were higher among individuals with coinfections. Males with ≥2 types had significantly higher titers for each type (except HPV 6) than males with single infections (P < 0.03 for 3 types); whereas, only HPV 11 titers were significantly higher among females with ≥2 types (P < 0.004). Among coinfected individuals, there was no association between titer levels for the types, i.e., a high titer for 1 type did not predict a high titer for another type, as might be expected if titer level was dependent on host-specific factors. However the analysis was limited by small numbers in some strata.
NATSAL II data indicated that age of first sex was lower for later birth-cohorts, with a more pronounced trend in females (Figure A1, Supplemental Digital Content 2, online only, available at: https://links.lww.com/OLQ/A13 demonstrates age of first sex trends). The method used to extrapolate age of first sex in the under 25 and over 45 age group did not alter which model was the best fit to seroprevalence data, except in 1 case (Table 2 and Table A1, Supplemental Digital Content 3, online only, available at: https://links.lww.com/OLQ/A14, details model results). Therefore, subsequent results presented here are from models assuming the observed trend in age of first sex continued. The fitted models of seroprevalence by HPV type and gender are shown in Figure 3, while the Akaike Information Criterion for each model is given in Table 2. In all cases except 1 (males, HPV 11), models including an age-varying FOI were a better fit than the corresponding model with a constant FOI. Models including a term for waning antibodies tended to be a worse fit than corresponding models without one. The models that best fit the data included terms for age-dependent FOI and cohort effects, but not a term for waning, although there were exceptions (HPV 16 in females, HPV 18 in males).
In this cross-sectional survey, we found type-specific seroprevalence increased from adolescence and then reduced in older females and plateaued in older males. We did not observe a decrease in antibody titers with increasing age (in either sex), while mathematical models suggested that age-dependent risk of infection, and not waning antibodies, best explained the observed seroprevalence trends.
Similar preimmunization trends have been reported in studies among females10,11 and males.10,11,21–23 Two of these have also found HPV 6 and 16 seroprevalence to be higher than that of HPV 11 and 18 among females.10,11 In a parallel study, we have found the prevalence of HPV 6 and 16 DNA to be over 3-fold that for HPV 11 and 18 DNA.24 It is unlikely that the differences are due to differing seroconversion rates as these have been reported to be similar for HPV 6, 16, and 18.9 Seroprevalence measures underestimate the cumulative HPV exposure among females because only 60% to 69% seroconvert25 and those with recent infections may not have seroconverted at the time of sampling, as median time between detection of HPV DNA and antibodies is 12 to 13 months.9 However, studies should be compared with caution because of differences in the populations studied and the assays used. Different assays have different sensitivities, which will impact the measurement of seroprevalence.
Seroprevalence was lower among young males than females for each HPV type. It is unlikely that the overall lower seroprevalence at all ages is due to lower infection rates in males, because they report at least as many lifetime partners as females, which increases their risk of exposure to HPV,26 they have rates of genital warts diagnoses similar to females,27,28 and overall HPV DNA prevalence studies have found prevalence to be comparable among males and females.29,30 It is more likely that differences lie in the immunologic responses produced by males. As infections in males are predominately in the epithelial surfaces, a detectable humoral immune response may be less likely to be induced than among females whose infections occur in mucosal surfaces. Transient infections are known to result in lower seroconversion rates in females,9 which may also be expected to contribute to low seropositivity in males. The proportion of males who fail to seroconvert is unknown. The ratio of female:male seropositivity of 2:1 in our study suggests—if all else is equal, and two-thirds of females seroconvert—that only around a third of males seroconvert (i.e., 64.5/2.2). In addition, males have been shown to serorevert after natural infection more often than females31; although we saw no evidence of declining antibody titer levels in males or females with age.
The lower seroprevalence observed by us, and some others, in older females has not been fully explained previously. Theoretical explanations for the decline seen in our survey include waning antibodies, birth cohort effects on the risk of infection, and age-specific biases in our sample. Seroprevalence surveys conducted in the United States and Australia among similar birth cohorts have found declining seroprevalence among women aged ≥40 years and have suggested that this decline may be attributable to waning antibodies, or birth cohort effects, or both.10,11 Studies in other countries, of varying birth cohorts and at varying ages, have also discussed sexual behavior changes and antibody waning as possible explanations for age-specific declines in seroprevalence.32–34
Our models that accounted for age-dependent changes in the force of HPV infection and for observed trends in age of first sex were best able to explain the pattern in seroprevalence. In contrast, incorporating waning of antibodies made little difference to the models' goodness of fit. Furthermore, GMT titers of seropositive individuals did not vary by age. Hence, both statistical and serological analyses support the hypothesis that a decline in antibody titers over time was not the primary explanation of the observed trend, whereas changes in risk within birth cohorts over the past 30 years were a key factor. Older females had a higher age of first sex, whereas the age of sexual debut in older males show much smaller changes. Sexual behavior has previously been shown to be a key determinant of seropositivity among females. HPV 16 and 18 seropositivity was higher among females who initiated sex before 16 years.32 The higher reported age of first sex in older females may be due to differential social desirability bias. However, the proportion of those in the same age cohort reporting sexual debut before 16 years in NATSAL II in comparison to NATSAL I has increased, suggesting that the age of first sex in females has lowered over time regardless of this bias.35 We have had to use age of first sex as a proxy for sexual risk taking in a particular age cohort but cannot be sure that this is the key determinant of risk. Although these data were not available on the youngest and oldest cohorts, making alternative assumptions about trends in these age groups did not alter models of best fit.
The population tested in this survey is not randomly selected but consists of anonymized residual serum samples submitted routinely for microbiologic diagnostic or screening investigation. Target numbers of sera are collected from the entire age range by region in England each year. Previous serological studies using the same collection found the sample source to be broadly representative of the general population.14 However, as our peak prevalence in females is found in the child-bearing age groups, there may be biases specific to our analysis due to a possible predominance of antenatal samples in these age-groups, which could not be corrected by modeling.
This study extends previously published findings on seroprevalence of young English females to older females and males and provides useful baseline epidemiologic data to monitor the impact of the HPV immunization programme on reducing HPV prevalence. Comparison with seroprevalence studies after widespread immunization will show changes in exposure to infection amongst unvaccinated men and women. It is the first study to combine seroprevalence data with mathematical models to explain the age distribution of seroprevalence. In addition, it is one of a small number of studies to provide estimates of age-dependent seroprevalence in males.
1. Clifford G, Franceschi S, Diaz M, et al. Chapter 3: HPV type-distribution in women with and without cervical neoplastic diseases. Vaccine 2006; 24:S3/26–S3/34.
2. Parkin DW, Bray F. Chapter 2: The burden of HPV-related cancers. Vaccine 2006; 24:S3/11–S3/25.
3. Lacey CJ, Lowndes CM, Shah KV. Chapter 4: Burden and management of non-cancerous HPV-related conditions: HPV-6/11 disease. Vaccine 2006; 24:S3/35–S3/41.
4. Department of Health. HPV vaccine recommended for NHS immunisation programme. Available at: http://www.immunisation.nhs.uk/Library/News/HPV_vaccine_Press_Release_26–10–07
. Accessed January 26, 2010.
5. Kitchener HC, Almonte M, Wheeler P, et al. HPV testing in routine cervical screening: Cross sectional data from the ARTISTIC trial. Br J Cancer 2006; 95:52–61.
6. Cuschieri KS, Cubie HA, Whitley MW, et al. Multiple high risk HPV infections are common in cervical neoplasia and young women in a cervical screening population. J Clin Pathol 2004; 57:68–72.
7. Sargent A, Bailey A, Almonte M, et al; ARTISTIC Study Group. Prevalence of type-specific HPV infection by age and grade of cervical cytology: Data from the ARTISTIC trial. Br J Cancer 2008; 98:1704–1709.
8. Moscicki AB, Schiffman M, Kjaer S, et al. Chapter 5: Updating the natural history of HPV and anogenital cancer. Vaccine 2006; 24:S3/42–S3/51.
9. Carter JJ, Koutsky LA, Hughes JP, et al. Comparison of human papillomavirus types 16, 18, and 6 capsid antibody responses following incident infection. J Infect Dis 2000; 181:1911–1919.
10. Markowitz LE, Sternberg M, Dunne EF, et al. Seroprevalence of human papillomavirus types 6, 11, 16 and 18 in the United States: National health and nutrition examination survey 2003–2004. J Infect Dis 2009; 200:1059–1067.
11. Newall AT, Brotherton JM, Quinn HE, et al. Population seroprevalence of human papillomavirus types 6, 11, 16, and 18 in men, women and children in Australia. Clin Infect Dis 2008; 46:1647–1655.
12. Chen CJ, Viscidi RP, Chuang CH, et al. Seroprevalence of human papillomavirus types 16 and 18 in the general population in Taiwan: Implication for optimal age of human papillomavirus vaccination. J Clin Virol 2007; 38:126–130.
13. Jit M, Vyse M, Borrow R, et al. Prevalence of human papillomavirus antibodies in young females subjects in England. Br J Cancer 2007; 97:989–991.
14. Osborne K, Gay N, Hesketh L, et al. Ten years of serological surveillance in England and Wales: Methods, results, implications and action. Int J Epidemiol 2000; 29:362–368.
15. Opalka D, Lachman CE, MacMullen SA, et al. Simultaneous quantitation of antibodies to neutralizing epitopes on virus-like particles for human papillomavirus types 6, 11, 16, and 18 by a multiplexed luminex assay. Clin Diagn Lab Immunol 2003; 10:108–115.
16. Dias D, Van Doren J, Schlottmann S, et al. Optimization and validation of a multiplexed luminex assay to quantify antibodies to neutralizing epitopes on human papillomarivuses 6, 11, 16 and 18. Clin Diagn Lab Immunol 2005; 12:959–969.
17. Muench H. Catalytic Models in Epidemiology. Cambridge, MA: Harvard University Press, 1959.
18. Jit M, Gay N, Soldan K, et al. Estimating progression rates for human papillomavirus infection from epidemiological data. Med Decis Making 2010; 30:84–98.
19. Wellings K, Nanchahal K, Macdowall W, et al. Sexual behaviour in Britain: Early heterosexual experience. Lancet 2001; 358:1843–1850.
20. Akaike H. A new look at the statistical model identification. IEEE Trans Automat Contr 1974; 19:716–723.
21. Dunne E, Nielson CM, Stone K, et al. Prevalence of HPV infection among men: A systematic review of the literature. J Infect Dis 2006; 194:1044–1057.
22. Kreimer AR, Alberg AJ, Viscidi R, et al. Gender differences in sexual biomarkers and behaviors associated with human papillomavirus -16, -18 and -33 seroprevalence. Sex Transm Dis 2004; 31:247–256.
23. Dunne EF, Nielson CM, Hagensee ME, et al. HPV 6/11, 16, 18 seroprevalence in men in two US cities. Sex Transm Dis 2009; 36:671–674.
24. Howell-Jones R, Bailey A, Beddows S, et al. Multi-site study of HPV type-specific prevalence in women with cervical cancer, intraepithelial neoplasia and normal cytology, in England. Br J Cancer 2010; 103:209–216.
25. Dillner J. The serological response to papillomaviruses. Semin Cancer Biol 1999; 9:423–430.
26. Svare EI, Kjaer SK, Nonnenmacher B, et al. Seroreactivity to human papillomavirus type 16 virus-like particles is lower in high-risk men than in high-risk women. J Infect Dis 1997; 176:876–883.
27. Pirotta M, Stein AN, Conway EL, et al. Genital warts incidence and health care resource utilisation in Australia. Sex Transm Infect 2010; 86:181–186. doi: 10.1136/sti.2009.040188.
28. Simms I, Fleming DM, Lowndes CM, et al. Surveillance of sexually transmitted diseases in general practice: A description of trends in the Royal College of General Practitioners Weekly Returns Service between 1994 and 2001. Int J STD AIDS 2006; 17:693–698.
29. Guiliano AR, Lu B, Nielson CM, et al. Age-specific prevalence, incidence, and duration of human papillomavirus infections in a cohort of 290 US men. J Infect Dis 2008; 198:827–835.
30. Kjaer S, Munk C, Winther JF, et al. Acquisition and persistence of human papillomavirus infection in younger men: A prospective follow-up study among Danish soldiers. Cancer Epidemiol Biomarkers Prev 2005; 14:1528–1533.
31. Thompson DL, Douglas JM, Foster M, et al. Seroepidemiology of infection with human papillomavirus 16 in men and women attending sexually transmitted disease clinics in the United States. J Infect Dis 2004; 190:1563–1574.
32. Wang SS, Schiffman M, Shields TS, et al. Seroprevalence of human papillomavris-16, -18, -31 and -45 in a population-based cohort of 10 000 women in Costa Rica. Br J Cancer 2003; 89:1248–1254.
33. Kramer M, Mollema L, Smits G, et al. Age-specific HPV seroprevalence among young females in the Netherlands. Sex Transm Infect 2010; 86:494–499. doi: 10.1136/sti.2009.041210.
34. af Geijersstam V, Wang Z, Lewensohn-Fuchs I, et al. Trends in seroprevalence of human papillomavirus type 16 among pregnant women in Stockholm, Sweden, during 1969–1989. Int J Cancer 1998; 76:341–344.
35. Copas AJ, Wellings K, Erens B. The accuracy of reported sensitive sexual behaviour in Britain: Exploring the extent of change 1990–2000. Sex Transm Infect 2002; 78:26–30.