This report of population-based seroprevalence of HPV among Chinese women, to our knowledge, is the first to be conducted. Within five different geographic regions across Western to Eastern China, we found a higher HPV seroprevalence in women from metropolitan areas, Uyghur ethnic groups, and in Shanghai city. Differences in sexual behavior, as well as differences in underlying population-based seroprevalence, are likely causes of differences in HPV seropositivity across regions. Overall seroprevalence of HPV 6, 11, 16 and 18 among Chinese women increased with age, and were strongly correlated with higher grades of cervical lesions.
A relatively low seroprevalence to HPV 16 and 18 was found among Chinese females, as compared to other areas worldwide . Given the large population base and relatively conservative sexual behavior among women in China within the past few decades, prevalence of serological responses to HPV is, as expected, to be low as compared to a relatively wide range of seroprevalence observed globally. Regarding Asia, HPV 16 and 18 seroprevalence in our study is most comparable to a population-based study in Taiwan (6.3% vs. 7.6%, 2.1% vs. 3.9%, respectively) , but much lower than that observed in Mongolia (6.3% vs. 23%, 2.1% vs. 19.6%) . HPV 16 seroprevalence is at similar levels to that in these five areas of China and Busan, South Korea (6.3% vs. 6.3%), whereas anti-HPV 18 was less prevalent than in South Korea (2.1% vs. 9.0%) . The seroprevalence data reported here is consistent with HPV genotyping data showing that HPV 16 is the dominant type measured by DNA in cervical samples in China [26, 27]. Seroprevalence of HPV 16 was found consistently to be more prevalent than HPV 18 in all areas of China. HPV 6 was the most prevalent type found in our study, which is consistent with other studies showing that HPV 6 is the most prevalent in cross-sectional studies . Because HPV 6 is not among the most common types of HPV detected by DNA cervical samples in China , we speculate a high probability of cumulative exposure to HPV 6 in China, although the specific reason is unknown. Some possibilities include higher clearance of HPV 6 than other types or lower induction or persistence rate of antibody of other HPV types upon their infection. There is evidence that incidence of genital warts has increased in China in recent years [29, 30], of which HPV 6 accounts for 54.9% .
The age trend of HPV seropositivity in our study was characterized by a peak for anti-HPV seroprevalence to any type in the late 40s, and a relatively flat curve for individual HPV types across different age groups. Such age-specific differences are possibly due to the fact that HPV seropositivity reflects cumulative exposure, which is higher among older women. High-risk HPV DNA prevalence peaks in both the early twenties and early forties in Chinese women  and low-risk HPV DNA generally has a very low prevalence in China. Peak HPV seroprevalence appeared after the second peak of DNA prevalence among Chinese women. It has been previously hypothesized that HPV viral load and persistence are the two most important predictors for HPV seroconversion . The sudden decline in seroprevalence in women in their late forties, which has also been reported in other studies [23, 33–35], might be due to waning immune response in older people or relatively lower exposure rates indicative of a cohort effect. Age-specific trends in HPV 16/18 seroprevalence were characterized by a steady increase trend across age, as compared to that in Mongolia , Thailand , Taiwan , and Costa Rica  where HPV 16/18 seroprevalence was relatively constant across age. Of note, the age range in which HPV 16 seroprevalence peaked in China was much later than in US (45-49 vs. 25-29), similar for HPV types 6, 11, and 18 . Overall, age-specific seroprevalence in China showed similar trend to other areas, with steady increases with age.
The association with HPV seropositivity and the subsequent risk of cervical neoplasia and cervical cancer remains inconsistent in natural history studies. However, several studies have shown positive associations between HPV seropositivity and cervical precancerous lesions [36–38]. Very few studies have reported a relationship between HPV seropositivity and the relative grade of cervical lesions. HPV seropositivity stratified by pathology diagnosis in our study showed that serological response to HPV 16 antigen was higher among women with CIN2 or above, but not among women with CIN1. This is likely due to the fact that persistent infection of HPV 16 is a prerequisite to the development of high-grade precancer or cancer. A higher increased risk of CIN3/cancer or CIN2/3 was found for both HPV 16 and 18 in Costa Rica  or in America , but not in the Czech Republic (CIN2/3) . In contrast to HPV 16, HPV 18 seropositivity was not associated with a higher risk of CIN2/3 in our study, possibly due to the relatively low observed HPV 18 seroprevalence. Seropositivity for HPV 11 was significantly higher in women with HSIL or greater ascertained cytology (OR = 4.1; 95% CI: 1.7-10.1, P < 0.001), but not by pathology diagnoses. This may be due to low HPV 11 seroprevalence or misclassified cytological diagnoses. Three of eight anti-HPV 11-positive patients with high grade cytology were diagnosed as normal or CIN1 by pathology. A similar pattern was not seen for anti-HPV 18 or 6.
Seroprevalence among virgins was low though detectable (0.6%), as compared to other studies which have reported a range from 0-3% of HPV 16 [33, 41, 42]. Given that HPV can be acquired at other non-genital sites, these findings may reflect possible non-intercourse sexual transmission . Further, sexual activity was self-reported by virgins in our survey, and therefore caution should be taken when interpreting these results. Women reporting a greater number of sexual partners have been found to have higher seroprevalence of HPV 16 [15, 44–46], 18 , and 11 , with many studies reporting a linear trend [15, 45, 46]. Among sexually active women in our study, there was no linear association with the percentage of seropositive samples and increasing number of reported sexual partners for any HPV type. Seropositivity to HPV 6 or 16 increased sharply from one to two partners, and slowly reached the plateau at more than four partners. This plateau phenomenon was also observed in a study from Norway , in which plateau of HPV seropositivity was reached at three or more partners.
The “male factor” is an important source of cumulative exposure of HPV to women. Though reports of husbands’ extramarital relationship were provided by wives via questionnaire and might be inaccurate, associations between HPV seropositivity and husbands’ sexual activity were still observed when husbands were reported to have extramarital sexual relationships. Induced abortion might be a marker of a greater number of pregnancies or number of lifetime sexual partners, thus explaining its association with higher anti-HPV 16 levels, results of which are consistent with a Mongolian study . Education level, smoking, drinking, pregnancy history, age at first menstruation or first sexual intercourse, and contraceptive use had no impact on seropositivity in our study, agreeing with the findings of several other studies [24, 35, 47], but not all .
The multiplexed Luminex immunoassay can simultaneously quantitate neutralizing antibodies to multiple HPV types in a large sample size, which is suitable for our study design . Cross-reactivity between HPV 6 and HPV 11 cannot be ruled out due to many conservative sequences shared by these two HPV types . Nevertheless, seroreactivity appears to be generally type-specific as the seroprevalence of HPV 6 was much higher than other three types. A previous Seattle cohort study of university women found a higher seroconversion rate for HPV 6 than HPV 16 or 18 or 11 or 45, following detection of the same or different type of HPV DNA [48, 49]. Considering time to seroconversion and antibody persistence, seroconversion to HPV 6 occurred earlier with same-type DNA detection than HPV 16, whereas anti-HPV 6 was not as persistent as anti-HPV 16 or 18 [48, 49]. Carter et al . studied HPV 16 and 18 antibody response following incident infections over years and found out that women in whom HPV DNA was detected at several visits may be significantly more likely to seroconvert than are women with only one HPV DNA-positive visit. Multiple HPV seropositivity in our study is not high between HPV 6 and 11 (1.3%) or 16 and 18 (0.6%) among sexually active women. On the other hand, no multiple seropositivity between HPV 6 and 16 or 18 and 16 was observed in women with a pathological grading of CIN3 or above, whereas the only two women seropositive for HPV 11 with CIN3 or above were also positive for anti-HPV 16. It might be because serum antibody response is a marker of lifetime cumulative exposure and some cross protection between types might exist . However, lack of infection and seroconversion history over time in this cross-sectional study cannot reinforce explanation of cross protection between specific types.
The strength of our study is that it was a multi-center study with a large sample size. The study covered five sites across China, including both rural and urban areas. We also included women with a wide age range from 15 to 54, stratified into five-year age groups. The number of virgins we recruited was also relatively large, providing source data for future planning of a vaccination program. Also, both the DNA and serology test used in this study are well validated and standardized. Among study limitations, data obtained from the five geographic sites might not be representative of the entire nation of China, particularly given the variation of HPV seropositivity across the sites or age groups. Also, women were recruited through posters or advertisement, but not by random sampling, indicating possible selection bias in our study. This bias is, to some extent, offset by the relatively large sample size, wide age range of participating women, as well as the inclusion of multiple geographical sites. Self-reported sexual behavior and their husbands’ extramarital sexual behavior by female subjects may also have introduced some bias.