Certain high-risk types of human papillomavirus (HPV), most notably HPV types 16 and 18, are the necessary cause of invasive cervical cancer (ICC) and its related precursors.1 The finding that nearly all ICC world-wide contain HPV DNA (99%)2, 3 has encouraged efforts to develop both prophylactic and therapeutic vaccines against HPV.4 Newly developed prophylactic vaccines based on HPV virus-like particles (VLPs) have been demonstrated to be safe and highly immunogenic in humans.5, 6
Before HPV vaccines may be introduced at the population-level, age-specific data on the prevalence of genital HPV infection in population-based samples of women are needed. The availability of such data is still limited,7, 8, 9, 10 particularly in Asia.11, 12 The prevalence of serum IgG antibodies against virus-like particles (anti-VLPs) of the most common HPV types, mainly HPV 16, has been evaluated in several populations.13, 14 Anti-VLPs are considered good markers of cumulative HPV exposure but are relatively insensitive markers of current HPV infection.13
In order to investigate whether differences in the age-specific prevalence of HPV infection in the female general population can explain the marked geographical variation in ICC incidence, the International Agency for Research on Cancer (IARC) coordinated a series of population-based prevalence surveys of HPV infection in areas with high and low incidence of ICC.15 As part of this effort, a prevalence survey of HPV infection was conducted in Busan, South Korea, an area with intermediate ICC incidence rates (20.8/100,000 in 1996–1997).16 Pap smear for the early detection of cervical cancer was first recommended in Korea in 1985 by the medical insurance of governmental employees and school teachers, but it has not been widely used so far nor included in any population-based screening program. This paper reports results on the prevalence of and determinants for detection of cervical HPV DNA and serum anti-VLPs among women in South Korea.
MATERIAL AND METHODS
A population-based survey was carried out between November 1999 and March 2000 in Busan, the second largest city of South Korea with an estimated population of 3.8 million people. A 2-stage cluster sampling method was used to randomly select female residents aged 15 years or older from the S district. In the first stage, 4 of 16 dongs of the S district were randomly selected. In the second stage, 3 sub-units of each dong (tongs) were randomly selected. A list of all female residents aged 15 years or older (N=2,684) from the 12 selected tongs was generated. Three hundred eighty-four women (14%) did not live at the given address, 260 (9.7%) refused to participate and 47 (1.5%) were not eligible due to exclusion criteria (non-Korean ethnicity, current pregnancy, history of previous hysterectomy or conization, and mental incompetence). An additional 313 women (11.7%) could not be contacted by interviewers. In total, 1,680 women were thus interviewed in their homes by 1 of 2 female nurses. The questionnaire included information on sociodemographic characteristics, smoking, drinking, Pap smear screening history, reproductive and menstrual factors and use of contraceptive methods. Following the interview, women were given an appointment for a medical examination in either the Dong-A or Kosin University Hospital. A total of 1,064 (63.3%) of the women interviewed at home came to the appointment. During the medical examination, a few questions on the sexual habits of the woman and her husband were asked. All participants signed informed consent forms according to the recommendations of the IARC and local ethical review committees who had approved the study.
Gynaecological examination and specimen collection
A total of 909 sexually active women accepted to undergo a routine gynaecological examination and to give 10 ml of blood. An additional 103 women who reported to be virgins accepted to provide a blood sample but not to have the gynaecological examination. Samples of exfoliated cells from the ectocervix and from the endocervix were collected with a cytobrush (Cervibrush, CellPath). After the preparation of a Papanicolau (Pap) smear, the remaining exfoliated cervical cells were placed in 50 ml conic tubes that contained 20 ml of phosphate-buffer solution (PBS) solution. All samples were sent daily to the laboratory of the Dong-A University Hospital for processing and storing. Samples of exfoliated cervical cells were centrifuged at 3,000g and the resulting pellet diluted in 1 ml of buffer solution and poured into labeled nunc tubes. Blood samples were centrifuged at 1500g for 10 min and aliquoted into plasma, buffy coats and red blood cells. All samples were stored in pre-labeled tubes at −70°C until shipment on dry ice to IARC.
HPV and HSV-2 detection techniques
HPV DNA test.
HPV testing was performed on exfoliated cervical cells from 894 women after excluding 15 women with inadequate cervical specimens. To analyze the quality of target DNA for polymerase chain reaction (PCR) purposes, cervical specimens were screened with β-globin gene-specific primers. After excluding samples from 31 women (3.5%) with β-globin negative samples, 863 cervical samples were tested for HPV DNA.
HPV positivity was assessed by DNA polymerase chain reaction-enzyme immunoassay (PCR-EIA) by using a HPV oligoprobe cocktail to detect the following 36 HPV types: HPV 16, 18, 31, 33, 35, 39, 45, 51, 52, 56, 58, 59, 66, 68, 6, 11, 26, 34, 40, 42, 43, 44, 53, 54, 55, 57, 61, 70, 71 (equivalent to CP8061), 72, 73, 81 (equivalent to CP8304), 82/IS39, 83 (equivalent to MM7), 84 (equivalent to MM8) and CP 6108.17 In addition, HPV positivity was assessed by low-stringency Southern blot analysis of PCR products with a cocktail probe of HPV-specific DNA fragments.18 Subsequently, GP5+/6+ PCR was repeated on positive samples in triplicate to generate sufficient products for further typing. After pooling these PCR products, typing was performed using PCR-EIA and HPV type-specific oligoprobes for the HPV types described above.19 Samples that were GP5/6+ positive by low-stringent Southern blot analyses but could not be identified by the above-mentioned EIAs were considered as HPV X, uncharacterized HPV types. Special precautions were taken to minimize false-positive results in the PCR, as has been described in detail elsewhere.18
HPV types considered as HR for this analysis included HPV types 16, 18, 31, 33, 35, 39, 45, 51, 52, 56, 58, 59, 66 and 68. The group of low-risk (LR) types included all other HPV types tested. Some of the low-risk HPV types are recognized as non-cancer associated types (e.g., HPV types 6, 11, 40, 42, 43 and 44), whereas others are HPV types with undetermined oncogenic potential (e.g., HPV types 26, 34, 53, 54, 70, 72 and 73). HPV infections having more than 1 HPV type were considered as HR if any HPV type detected was a HR type.
HPV VLP ELISA.
Serological testing for detection of IgG antibodies against HPV VLPs was conducted using an ELISA described previously.14 HPV VLPs were produced in Sf21 insect cells using recombinant baculoviruses encoding the L1 gene of HPV 16, 18, 31, 33 and 5814, 20, 21 and purified by isopycnic ultracentrifugation. Flat-bottomed wells of 96-well microplates (Maxisorp, Nunc, Life Technologies, Eragny, France) were coated overnight at 4°C with 200 to 800 ng of VLPs (test well) or 200 ng of BSA (control well) in PBS, pH 7.4. Each plasma sample was tested at a dilution of 1/10 against each of the 5 VLP types and BSA. After washing with PBS, 0.1% Tween 20 and 200 μl of PBS containing 1% newborn bovine serum (NBS, Sigma Chemical Co., St. Quentin Favallier, France) were added. The blocking solution was replaced by 100 μl of plasma diluted 1:10 in 5 × PBS-10% NBS and 2% Tween 20, and plates were incubated at 45°C for 60 min. After 4 washes, bound antibodies were detected with a goat anti-human IgG immunoglobulin (diluted 1:5,000) conjugated to horseradish peroxidase (Sigma Chemical Co.). Following incubation at 45°C for 1 hr and 4 washes, 100 μl of a substrate solution containing ortho-phenylenediamine and H2O2 was added. After a 30 min incubation, the reaction was stopped by addition of H2SO4 and optical densities (OD) were read at 492 nm. For each serum sample the background reactivity found in the BSA coated wells was subtracted from the OD found in each of the HPV-VLP coated wells. Negative values were adjusted to zero. The cut-off values for positivity were set up at 0.4.
HPV VLP findings were available for 860 women with information on cervical HPV DNA and Pap smear, and on 103 virgin women who did not undergo a gynaecological examination.
HSV-2 serum antibodies.
The presence of type-specific plasma IgG antibodies against HSV-2 were tested blindly using a HSV-2 ELISA assay developed by Focus Technology/formerly MRL (Cypress, CA).23 All HSV-2 positive findings were confirmed by means of a second test.
Odds ratios (ORs) for the detection of either HPV DNA or anti-VLPs and corresponding 95% confidence intervals (CIs) were calculated by means of unconditional, multiple logistic regression equations, adjusted for age (< 35; 35–44; 45–54; ≥ 55 years). All variables that showed statistically significant associations with either of the HPV markers in the age-adjusted analyses were finally included in the same model. The statistical significance of trends for ORs was assessed by considering the categorical variable as a continuous variable in the logistic model.
Pap smear findings were classified according to the Bethesda system by an experienced cytologist in the Department of Pathology at the Dong-A University Hospital. They included 866 (95.3%) smears with normal findings, 19 (2.1%) with atypical squamous cells of undetermined significance (ASCUS), including 1 with atypical glandular cells (AGUS), 17 (1.9%) low-grade squamous intraepithelial lesions (LSIL), and 4 (0.4%) high-grade squamous intraepithelial lesions (HSIL). One woman (0.1%) was diagnosed as carcinoma in situ (CIS) and 2 (0.2%) were diagnosed with ICC. Women with abnormal cytological findings underwent a colposcopical examination, confirmatory biopsies and treatment, when necessary.
Prevalence and type-specific distribution of HPV DNA and anti-VLPs
The prevalence of HPV DNA by HPV type and cytological findings was analyzed among 863 women with a β-globin positive sample (Table I). Overall, 90 women (10.4%; 95% CI: 8.5–12.7) were positive for HPV DNA. Prevalences standardized by age on the basis of the female population of Busan and of the world standard population were 12.9% and 13.0%, respectively. On account of multiple infections, 115 infections with 29 different HPV types were detected. Single infections (73/90 women) and infections with HR HPV types (54/90 women, 66/115 infections) predominated slightly. HPV prevalence was 8.5% among cytologically normal women, 22.2% among those with ASCUS, 52.9% among those with LSIL and 100% among 7 women with HSIL or worse (Table I). The most often found HPV types, in either single or multiple infections, were HPV 70 (14 women), HPV 16 (11 women) and HPV 33 (10 women), but the type distribution varied by cytological findings. A HR HPV type was involved in all 16 women with LSIL or worse, but in 36/70 of cytologically normal women. Among the latter, the most frequent HPV types were HPV 70 (13 women), HPV 33 (9 women) and CP 8304 (7 women).
Table I. Prevalence of Various Human Papillomavirus (HPV) Types by Cytological Findings in 863 Women in Busan, South Korea, 1999–20011
The prevalence of anti-VLPs against 5 HPV types among 860 sexually active women was evaluated according to the presence of HPV DNA of the corresponding or different type and by cytological findings (Table II). Overall, anti-VLPs were identified among 170 women (19.8%; 95% CI: 17.2–22.6). The most frequently found anti-VLPs was anti-HPV 18 (9.0%), followed by anti-HPV 31 (7.6%) and anti-HPV 16 (6.3%). Fourteen of the 30 (46.7%) women who were HPV DNA positive for any of the 5 types considered were anti-VLP positive, while 14/170 (8.2%) women with anti-VLPs were HPV DNA positive. Thus, concordance between HPV DNA and anti-VLPs in individual women was modest but significant for HPV 16 (OR of having anti-VLPs in the presence of HPV DNA = 9.8), HPV 31 (OR = ∞), HPV 33 (OR = 10.6), HPV 58 (OR = 15.1) and for the combination of any of the 5 types (OR = 3.8). Concordance was poor for HPV 18: none of 77 women with anti-HPV 18 VLPs were positive for HPV 18 DNA (Table II).
Table II. Concordance Between Cervical HPV DNA and Serum Anti-VLPS Among 860 Women in Busan, South Korea, 1999–20011
Anti-VLPs against any HPV type were found only slightly more often among women with abnormal cytology (26.2%) than among those with normal findings (19.4%) (OR = 1.5, non-significant).
Among 103 virgin women, 4.9% had anti-VLPs. Below age 25, only one of 73 virgin women had anti-VLPs (against HPV 18) (not shown). Virgin women are not included in the analyses that follow.
Figure 1 shows age-specific incidence rates of ICC in Busan and age-specific prevalence of detection of HPV DNA of any type, DNA of and anti-VLPs against HPV types 16, 18, 31, 33 and 58, and anti-HSV 2 in study women. Incidence rates for cervical cancer increased with age, up to 50–54 years, declined at age 55–59 and then increased again above age 60. Two (28.6%) of 7 women below age 25 had HPV DNA (types 66 and 68) (Table III). The prevalence of HPV DNA was approximately 10% between 25 and 60 years of age, but 4% among women aged 60 or older. All 7 sexually active women under 25 years of age were negative for antibodies against HPV 16, 18, 31, 33 and 58 (Table III). Seropositivity for anti-VLPs was approximately 25% in the 25–54 age range and declined thereafter. The age-curve for HPV DNA of the same types as anti-VLPs was similar to, although lower than, the anti-VLPs age-curve. Seropositivity for anti-HSV-2 increased steadily with age and became, after age 35–39, 2-fold higher or more than seropositivity to anti-VLPs.
Table III. Detection of Cervical HPV DNA and Serum Anti-HPV VLPS According to Socio-Demographic Characteristics. Busan, South Korea, 1999–20011
DNA from HR HPV types was approximately 4-fold more common than DNA from LR HPV types below age 35, whereas LR HPV types were approximately equally frequent as HR ones in women aged 35 or older. Cervical single HPV infections were more frequent than multiple infections in all age groups. Overall, 69% of anti-VLP positive women had anti-VLPs against one HPV type only (not shown).
Risk factors for HPV and anti-VLPs
Age-adjusted associations between various factors and the detection of HPV DNA and anti-VLPs are shown in Tables III–V. Differences by age group did not attain statistical significance for either of the 2 HPV markers considered, whereas years of education were inversely related to anti-VLPs (OR ≥ 13 vs. < 9 = 0.5) but not HPV DNA (Table III). Women who drank alcoholic beverages 1–2 times per week and smoked had an increased risk of detection of HPV DNA (OR = 2.0 and 2.4, respectively) and anti-VLPs (OR = 1.7 and 2.2, respectively) compared to non-drinkers and non-smokers. Divorced women, widows and, most strongly, single women had an excess of HPV DNA and anti-VLP detection compared to married women, whereas history of pap smear (72% of study women) was unrelated to either of the 2 HPV markers.
Table IV. Detection of Cervical HPV DNA and Serum Anti-HPV VLPS by Menstrual and Reproductive Characteristics. Busan, South Korea, 1999–2001
Among menstrual and reproductive characteristics, age at menarche was inversely related to HPV DNA (OR for ≥ 18 vs. ≤ 15 years = 0.5) but not anti-VLPs (Table IV). The detection of either HPV marker was lower among parous women (OR for ≥ 3 births = 0.4 for HPV DNA and 0.3 for anti-VLPs) than among nulliparous women. Age at first marriage or first pregnancy, number of induced or spontaneous abortions and menopausal status were unrelated to HPV markers (Table IV).
Both HPV markers were strongly associated with a woman's lifetime number of sexual partners (OR for ≥ 4 vs. 1 sexual partner = 3.5 for HPV DNA and 5.4 for anti-VLPs) and with seropositivity for anti-HSV 2 (ORs = 2.6 and 2.5, respectively) (Table V). Age at first intercourse seemed to have little influence on either HPV marker. Husband's extra-marital affairs and history of vasectomy were associated with detection of HPV DNA (ORs = 3.1 and 2.5, respectively) but not of anti-VLPs. Women who said to be uncertain about their husband's extra-marital affairs did not show an increased risk for either HPV marker. No clear risk pattern emerged in respect to the use of female contraceptive methods, including oral contraceptives and condoms (which were each used by 5% of study women) and intra-uterine devices (IUD) and tubal ligation (each reported by approximately one third of study women) (Table V).
When the variables that were most strongly related to HPV DNA and/or anti-VLP detection were entered in the same model (Table VI), all associations were slightly weakened, most notably those with smoking and a woman's sexual habits. Anti-HSV-2 remained significantly associated with HPV DNA and anti-VLP detection, whereas husband's extra-marital affairs and history of vasectomy were still related to HPV DNA but not anti-VLP detection.
Table VI. Major Risk Factors for Detection of Cervical HPV DNA and Anti-HPV VLPS. Busan, South Korea, 1999–20011
Analyses in Table VI were repeated separately in women younger than 45 and those aged 45 or older. No differences in the role of major risk factors for HPV DNA or anti-VLP detection were found.
The prevalence of cervical HPV DNA among sexually active women aged 15 years or older from Busan, South Korea, was 10.4%. Standardization by age on the basis of the female population of Busan or on the world standard population produced prevalence estimates of 12.9% and 13.0%, respectively. In agreement with previous work,7, 9, the majority of HPV cervical infection among young women involved HR HPV types, whereas above 35 years of age HR and LR types were at least equally frequent. Half of HPV-DNA positive women were infected with HPV 70, 16 and 33, in decreasing order. It is the first time HPV 70 is found to be relatively common in a population of cytologically normal women. Worldwide, only 22 out of 85 studies that showed the distribution of various HPV types in cervical cancer specimens using PCR-based assays reported on the prevalence of HPV 70.23 This was greater than 1% in 2 studies, including 1 from China.24
As expected,13, a larger proportion of sexually active women (19.8%) were positive for anti-VLPs than for HPV DNA. Standardized proportions for anti-VLPs on the basis of the female population of Busan or of the world standard population were 17.5% and 17.1%, respectively. Fourteen of the 30 women positive for HPV DNA types 16, 18, 31, 33 or 58 showed antibodies against any of these high-risk types, but the concordance between the markers for each specific HPV type was difficult to evaluate on account of low numbers of HPV DNA positive women. Some cross-reactivity between genital HPV types is also possible.13, 14 In the comparison of the findings from cervical measurement and serology, it is important to bear in mind that whereas cervical HPV DNA measurements were available for 34 HPV types, anti-VLPs reflected seroconversion vs. 5 HR HPV types only.
Few population-based surveys of the prevalence of HPV DNA in unselected female populations are available world-wide.7, 8, 9, 10, 11, 12, 17, 25 The findings from South Korea support the presence of a good correlation between the prevalence of HPV and incidence rates of ICC, at least in populations where screening programs do not exist or have been introduced very recently, as in South Korea. HPV prevalence in Busan is similar to those found in intermediate-risk developing countries, such as Lampang, Thailand11 and Ho Chi Minh City, Vietnam,12 where ICC incidence rates in the 20–25/100,000 women-range can be found,26 but lower than in higher-risk countries such as Costa Rica,7 Colombia,25 Argentina10 and Mozambique.27 A higher HPV DNA prevalence at a population level may also be correlated with a greater proportion of infection with multiple HPV types.7, 9 In the comparison of overall HPV prevalence and the prevalence of multiple HPV infections between studies, it is however important to bear in mind that different sampling procedures and PCR methods for HPV testing and typing had been used.7, 27
With respect to age-specific prevalence of HPV DNA, our survey provided very little information on women below 25 years because few women is this age group in Busan had already married, and unmarried women were generally not willing to undergo a gynaecological examination. HPV prevalence seemed, however, remarkably stable in the 30–59 age range. Our findings did not confirm, therefore, the existence of a second peri- or post-menopausal peak as seen in some Latin-American populations.7
It is the first time that the prevalence of HPV DNA in the cervix and of selected anti-VLPs can be compared in a large population-based sample of women from Asia. Some agreement between the detection of type-specific HPV DNA and anti-VLPs, at an individual level, was found, but it was modest. In particular, the highest seroprevalence was for HPV 18 (9.0%), yet only 1 woman was positive for HPV 18 DNA. Differences between HPV DNA and anti-VLPs, however, are not surprising as the 2 markers reflect different outcomes of HPV infection. Whereas HPV DNA prevalence in one point in time measures recent exposure and a subset of old exposures that have become persistent, anti-VLPs represent a marker of cumulative HPV exposure.13 None of the 7 sexually active women below age 25 was positive for either DNA of or antibodies against HPV 16, 18, 31, 33 or 58. The overall prevalence of anti-VLPs was similar in the 25–54 year range. Seropositivity for HPV thus showed a different age-pattern than seropositivity for HSV 2 that increased steadily with age and reached much higher levels than anti-VLPs (50% of middle-aged women).
Despite the differences above, risk factors for the detection of HPV DNA and anti-VLPs were remarkably similar. As in previous studies on risk factors for HPV DNA detection in cancer-free women,10, 20, 28, 29, 30 we found consistent associations between the number of sexual partners, seropositivity for HSV-2, and being single or divorced, and the positivity for either HPV marker. Similar direct associations were also found with alcohol drinking and smoking, although these were explained in part by differences in sexual habits. Previous findings on the role of smoking on HPV infection have been inconsistent, showing direct,12 no30 or even inverse, associations.11 In agreement with previous work, HPV infection, expressed as either HPV DNA or anti-VLP positivity, was inversely related to the number of births.11, 28
It is well known that in many populations, including probably South Korea, husbands' sexual habits are at least as much, if not more, important than women's sexual habits as a risk factor for sexually transmitted infections and, eventually, ICC.31 We had some information on the “male factor” and found an approximately 3-fold increased risk for HPV DNA detection among women who reported that their husband had extra-marital affairs or underwent vasectomy. In our study, vasectomy was not correlated to available indicators of sexual activity in men or corresponding wives and therefore this finding is difficult to interpret and needs to be confirmed in other populations. Interestingly, a history of extramarital affairs was not associated with seropositivity for anti-VLPs in the wife, as if a husband's sexual activity were a source of frequent re-infection with HPV, and therefore recent new cervical infections without seroconversion. The information on extra-marital affairs is however likely to be rather inaccurate and must be interpreted cautiously.
All the risk factors reported above seemed similarly important in women below and older than 45 years. No clear difference emerged in risk factors for cervical infection with HR and LR HPV types.
Several studies14, 32, 33 have shown that genital infection with HPV is followed by serologic response to HPV VLPs, but a substantial proportion of HPV-infected women fail to seroconvert. Furthermore, it is unclear how long anti-VLPs persist after seroconversion. Our serological data suggest anti-VLPs behave very differently from anti-HSV-2, the other sexually-transmitted disease we had measured. Late (L) HPV proteins are localized in the most superficial layer of cervical epithelium, relatively far away from blood vessels.34 Furthermore, HPV has a battery of immune-evasion mechanisms.35 Papillomavirus L genes, which encode the capsid proteins, contain codons that mammalian cells rarely use. The production of abundant papillomavirus capsid proteins (against which anti-VLPs are targeted) is inhibited in mammalian cells by the restricted availability of the appropriate t RNAs. Limited accessibility and production of L proteins, combined with many other transcriptional and translational control mechanisms, ensure limited opportunities for the host's immune system to respond to L proteins.34, 35
The present findings should be representative of the prevalence of HPV DNA and anti-VLPs in the population under study. One report on 130 cytologically normal women who visited a gynaecological department in Inchon, South Korea, also showed a prevalence of 10% for HPV DNA.36 The 2 major causes of non-participation in our survey were inaccuracies in the addresses in the resident list (approximately one fourth of randomly chosen women) and failure of women to come for the gynaecological examination (approximately one third of those who had an interview at their home). The refusal of many women to undergo cytological examination is of concern, although not surprising on account of the relative novelty of cytological cervical screening in Korea. A comparison of the characteristics of women who underwent the gynaecological examination and those who did not, showed, however, that the 2 groups were similar in respect to educational years, lifestyle habits, pap smear history and reproductive and menstrual factors.
In conclusion, the present survey provided a special opportunity to evaluate jointly the prevalence of HPV DNA and anti-VLPs. The vast majority of women below age 25 years in Busan were not married, reported to be virgins and did not show markers of infection with HPV 16 and 18. Pending confirmation in a larger survey of young women, it thus seems that a prophylactic vaccine against these 2 types, when available,37 may also be given in South Korea to women in their early twenties.
We thank the staff of the Department of Health and Hygiene, Busan City, and the Health Center of Seo-Gu, Busan, for their collaboration, and Dr. G. Clifford, Mrs. A. Arslan, Mr. Y. Guy, Ms. S. Dardeau and Ms A. Zangarelli for technical assistance. Dr. Rhoda Ashley conducted the testing for anti-HSV 2 serum antibodies. This paper was prepared during the tenure of a Technical Transfer Award of HR Shin at the International Agency for Research on Cancer.