High-risk (HR) human papillomavirus (HPV) prevalence has been shown to correlate well with cervical cancer incidence rates. Our study aimed to estimate the prevalence of HR-HPV and cervical intraepithelial neoplasia (CIN) in China and indirectly informs on the cervical cancer burden in the country. A total of 30,207 women from 17 population-based studies throughout China were included. All women received HPV DNA testing (HC2, Qiagen, Gaithersburg, MD), visual inspection with acetic acid and liquid-based cytology. Women positive for any test received colposcopy-directed or four-quadrant biopsies. A total of 29,579 women had HR-HPV testing results, of whom 28,761 had biopsy confirmed (9,019, 31.4%) or assumed (19,742, 68.6%) final diagnosis. Overall crude HR-HPV prevalence was 17.7%. HR-HPV prevalence was similar in rural and urban areas but showed dips in different age groups: at age 25–29 (11.3%) in rural and at age 35–39 (11.3%) in urban women. In rural and urban women, age-standardized CIN2 prevalence was 1.5% [95% confidence interval (CI): 1.4–1.6%] and 0.7% (95% CI: 0.7–0.8%) and CIN3+ prevalence was 1.2% (95% CI: 1.2–1.3%) and 0.6% (95% CI: 0.5–0.7%), respectively. Prevalence of CIN3+ as a percentage of either all women or HR-HPV-positive women steadily increased with age, peaking in 45- to 49-year-old women. High prevalence of HR-HPV and CIN3+ was detected in both rural and urban China. The steady rise of CIN3+ up to the age group of 45–49 is attributable to lack of lesion removal through screening. Our findings document the inadequacy of current screening in China while indirectly raising the possibility that the cervical cancer burden in China is underreported.
Cervical cancer is the third most common cancer in women with an estimated 530,000 new cases and 275,000 deaths worldwide per year.1 Corresponding estimates for the People's Republic of China are ∼75,500 new cases (14% of all cervical cancer worldwide) and 34,000 deaths (12%).1, 2 Despite these substantial disease numbers, the age-standardized cervical cancer incidence rate in China (9.6 per 100,000 women) is lower than the world average (15.2 per 100,000 women).1, 3 However, estimates of cervical cancer incidence in China derive from incidence–mortality ratios based on cancer registries.1 These cancer registries include less than 2% of the Chinese population and are located in urban areas.4 Conversely, 70% of the Chinese population reside in rural areas,5 where 90% of incident cervical cancer cases are estimated to occur.6
High-risk human papillomavirus (HR-HPV) is the necessary cause of precancerous cervical intraepithelial neoplasia (CIN) and invasive cervical cancer.7 HR-HPV prevalence has been found to correlate well with cervical cancer risk in corresponding populations, particularly in middle age women.8, 9 In addition, the positive correlation between HPV prevalence and the age-standardized incidence of cervical cancer by subcontinent has been reported recently.10 China's rapid industrialization and urbanization during the last three decades have coincided with an epidemic of sexually transmitted disease,11 which may correlate with an increase in HR-HPV and CIN prevalence as well.
Aiming to determine more realistic rates of HR-HPV, CIN and cervical cancer prevalence in China, we performed a pooled analysis on individual data from 10 years of cervical cancer screening projects, which included 30,207 women from several rural and urban areas throughout China.
Material and Methods
From 1999 to 2008, the Cancer Institute/Hospital of the Chinese Academy of Medical Sciences (CICAMS; Beijing, China), in collaboration with the Cleveland Clinic (Cleveland, OH), International Agency for Research on Cancer (IARC; Lyon, France) and Program for Appropriate Technology in Health (PATH; Seattle, WA), screened 30,207 women in population-based, cross-sectional cervical cancer screening studies in five urban and nine rural areas of China, throughout nine Han majority provinces. Eligible women were sexually active, not pregnant, had an intact uterus and had no history of treatment on the cervix. Recruitments for all the studies were based on community lists to minimize the selection bias. Most study women had never been screened for cervical cancer, and none had been screened in the last 5 years. Informed consent was obtained from all women. The study methodologies have been outlined in detail in the previously published works.12–19 The human subjects review boards of CICAMS and Cleveland Clinic or PATH or IARC approved these studies.
Information about individual studies is listed in Table 1 and has been published recently in great detail.19
Table 1. Characteristics of pooled studies
All participants underwent liquid-based cytology (LBC; SurePath™, BD Diagnostics, Franklin Lakes, NJ or ThinPrep®, Hologic, Bedford, MA), visual inspection with acetic acid (VIA) and HR-HPV DNA testing (HC2, Qiagen, Gaithersburg, MD). All women in Shanxi Province Cervical Cancer Screening Study (SPOCCS) I and those positive for any of the three screening tests in the other studies received biopsy performed using a 2-mm bronchoscopy biopsy instrument. Colposcopy-directed biopsy was used in cases of visible lesions. When study protocol included random (four-quadrant) punch biopsies (Table 1), biopsies were taken at positions of two, four, eight and ten o'clock at the squamocolumnar junction. Details about study procedure have been published in detail in prior reports.19
HPV DNA testing was performed using the HR probe set of Hybrid Capture 2 (HC2, Qiagen), which detects 13 carcinogenic HPV types (HPV 16, 18, 31, 33, 35, 39, 45, 51, 52, 56, 58, 59 and 68). All women received physician-collected HPV DNA sampling; for primary analyses, HPV DNA positivity was defined according to the manufacturer's recommended positive cut-point of 1.0 relative light units per cutoff (RLU/Co)(≈pg DNA/mL), as previously described.12, 20
Verification of disease status
In all studies, laboratory personnel performing HC2 were blinded to other test results, and cytopathologists and histopathologists made diagnoses without knowledge of other test results. Colposcopists were blinded to the results of all screening tests but were aware one test was positive, except in SPOCCS I, in which all women received colposcopy regardless of test positivity.12 In most studies, cytology and biopsy results were read at CICAMS, although some biopsies were read by local pathologists. Cytology results in eight studies and biopsy results in ten studies were reviewed for quality control by international experts (Table 1). In SPOCCS I and the Screening Technologies to Advance Rapid Testing Project studies, all abnormal pathological slides and 10% of the negatives were reviewed by international clinical experts for reinterpretation. Comparatively, in SPOCCS III studies, 10–35% abnormal pathological slides received international review, whereas in the IARC Yangcheng study, all abnormal cytological slides and 20% of the negatives were reviewed by international experts.
Data were pooled using a uniform disease-status classification based on histologically confirmed biopsies. Women without biopsy results, but with negative HR-HPV DNA and negative or atypical squamous cells of undetermined significance (ASC-US) at LBC results were considered to be true negatives, based on findings from SPOCCS I that found only one CIN2 case (1/1,511, 0.07% of population) and no CIN3+ cases among those women.12 Women without biopsy but with negative cytology, positive HR-HPV DNA and negative colposcopy were also categorized as negative. Women were considered as having incomplete data and excluded from the analysis if they had no biopsy but fell in one of these categories: (i) ASC-US and HR-HPV positive; (ii) low-grade squamous intraepithelial lesion or worse [including atypical squamous cells—cannot exclude high-grade squamous intraepithelial lesion (ASC-H), low-grade squamous intraepithelial lesion (LSIL), high-grade squamous intraepithelial lesion, atypical glandular cells of uncertain significance, adenocarcinoma in situ, adenocarcinoma and squamous cell carcinoma]; (iii) negative cytology, HR-HPV positive and missing/positive colposcopy and (iv) missing cytology.
Individual raw data from 17 studies were combined to estimate the prevalence of HR-HPV, different grades of CIN and cervical cancer. Additionally, overall age-standardized prevalence was calculated using the world standard population reported by Doll et al.21 Crude prevalence is reported when not otherwise specified. Prevalence stratified by age group (15–24, 25–29, 30–34, 35–39, 40–44, 45–49 and 50–59 years old) was also calculated for HR-HPV and CIN1, 2 and 3 or worse (CIN3+), and Cochran–Armitage trend test was used to test the trend of age-stratified prevalence. Additionally, logistic regression model was used to explore the effects of age and geographic location on HPV infection status and to explore whether international quality control affected CIN2+ prevalence, adjusting for age and geographic location. Age-stratified CIN prevalence in all women and in HR-HPV-positive women was restricted to rural women on account of the relatively low number of urban women in individual age groups.
ASC-US: atypical squamous cells of undetermined significance; >ASC-US: worse than atypical squamous cells of undetermined significance; CI: confidence interval; CICAMS: Cancer Institute/Hospital of the Chinese Academy of Medical Sciences; CIN: cervical intraepithelial neoplasia; CIN1: cervical intraepithelial neoplasia grade 1; CIN2: cervical intraepithelial neoplasia grade 2; CIN2+: cervical intraepithelial neoplasia grade 2 or worse; CIN3: cervical intraepithelial neoplasia grade 3; CIN3+: cervical intraepithelial neoplasia grade 3 or worse; ECC: endocervical curettage; HC2: digene HPV HC2 DNA test; HR-HPV: high-risk human papillomavirus; IARC: International Agency for Research on Cancer; LBC: liquid-based cytology; LSIL: low-grade squamous intraepithelial lesion; LSIL+: low-grade squamous intraepithelial lesion or worse; PATH: Program for Appropriate Technology in Health; SPOCCS: Shanxi Province Cervical Cancer Screening Study; VIA: visual inspection with acetic acid; VILI: visual inspection with Lugol's iodine
In total, 30,207 women were screened. Figure 1 describes the process by which we obtained our final sample of 29,579 women with HR-HPV testing results, of which 28,761 had biopsy confirmed (31.4%) or assumed (68.6%) final diagnosis. Most women (88.4%) in our study population were from rural areas. The average age of rural women was 40.2 years [standard deviation (SD): 6.2, range: 17–59 years]: most rural women were aged 35–39 (31.3%), 40–44 (28.5%) or 45–49 years (18.8%), though rural women aged 17–24 (1.6%), 25–29 (1.7%), 30–34 (12.2%) and 50–59 years (5.9%) were also included. Among rural women in our study population, almost all were married (98.7%), with a median number of three pregnancies [range: 0–13, Interquartile range (IQR): 2–4] and two live births (range: 0–8, IRQ: 2–3); the mean age of first childbirth was 22.0 years (SD: 2.81 years).
The remainder of our study population (11.6%) consisted of urban women. Their average age was 37.9 years (SD: 9.9, range: 17–59 years); 9.2% of urban women aged 17–24 years; 16.1% 25–29; 15.9% 30–34; 14.6% 35–39; 15.3% 40–44; 13.2% 45–49 and 15.7% aged 50 or older. The majority of urban women were married (89.7%), with a median of two pregnancies (range: 0–11, IQR: 1–3) and one live birth (range: 0–5, IQR: 1–1). The mean age at first childbirth among urban women was 26.0 years (SD: 3.28 years).
Sexual behavior differed by geographic location: rural women had a younger mean age of sexual debut (20.8 years, SD: 2.2 years) compared to urban women (23.7 years, SD: 3.1 years). Similarly, 22.1% of rural women reported having had two or more lifetime sexual partners compared to 15.7% of urban women. The mean age and sexual debut age of women excluded from the analysis were 39.1 (SD: 7.1) and 19.5 (SD: 2.6), respectively. The excluded population had the same median number of pregnancies and live births as included women.
HR-HPV DNA prevalence by geographic location
Overall crude and age-standardized HR-HPV prevalence in our study population were 17.7 and 16.8%, respectively (Table 2). When stratified by geographic location, crude and age-standardized HR-HPV prevalence were 18.0 and 16.3% [95% confidence interval (CI): 16.0–16.6%], respectively, in rural women and 15.2 and 16.0% (95% CI: 15.7–16.3%), respectively, in urban women (Table 2). When stratified by age, crude HR-HPV prevalence in rural women declined from 16.2% in 15- to 24-year-old women to 11.3% in 25- to 29-year-old women before increasing in women aged 35–39 years (18.6%) and then remaining stable (Ztrend = 5.079, ptrend < 0.0001; Fig. 2). Among urban women, the crude HR-HPV prevalence declined from 18.7% in 15- to 24-year-old women to 11.3% in 35- to 39-year-old women before increasing again to ∼16.0% in women aged 40 or older (Ztrend = −0.986, ptrend = 0.324; Fig. 2). Logistic regression analysis also showed that compared to the women aged 15–24 years, those aged 25–29 (OR = 0.769, OR 95% CI: 0.593–0.999) and 30–34 years (OR = 0.752, OR 95% CI: 0.609–0.929) had lower HPV prevalence, whereas women older than 35 years had a higher HPV prevalence comparable to that of women aged 15–24 years. This analysis also showed that HPV prevalence was lower in urban areas than in rural areas (OR = 0.870, OR 95% CI = 0.785–0.965).
Table 2. Crude and age-standardized prevalence of HPV infection and CINs by areas among all women and HPV-positive women
CIN1, CIN2 and CIN3+ prevalence
Overall crude and age-standardized prevalence were 3.4 and 3.1% (95% CI: 2.9–3.2%), respectively, for CIN1, 1.5 and 1.3% (95% CI: 1.2–1.3%), respectively, for CIN2 and 1.6 and 1.2% (95% CI: 1.1–1.2%), respectively, for CIN3+ (Table 2). Forty-nine women (48 rural and one urban) were diagnosed with cervical cancer (Table 2). Among rural women, age-standardized prevalence in rural women was 3.4% (95% CI: 3.2–3.5%) for CIN1, 1.5% (95% CI: 1.4–1.6%) for CIN2 and 1.2% (95% CI: 1.2–1.3%) for CIN3+. Lower age-standardized prevalence rates of CIN1 (2.1%, 95% CI: 2.0–2.2%), CIN2 (0.7%, 95% CI: 0.7–0.8%) and CIN3+ (0.6, 95% CI: 0.5–0.7%) were found in urban women (Table 2).
Age-stratified prevalence of CIN1, 2 and 3+ in rural women is shown in Figure 3. CIN1 prevalence was lowest in women younger than age 35 but increased in 35- to 39-year-old women (3.9%) and remained constant afterward (Ztrend = 0.710, ptrend = 0.478; Fig. 3). CIN2 prevalence in rural women was rather constant, with the highest level (1.9%) in 40- to 44-year-old women (Ztrend = 1.350, ptrend = 0.177). CIN3+ prevalence steadily increased up to 2.5% in 45- to 49-year-old women before slightly diminishing to 1.7% in women aged 50–59 years (Ztrend = 5.803, ptrend < 0.0001; Fig. 3).
After adjusting for age and geographic location, logistic regression analysis showed that CIN2+ prevalence was lower in studies that had histological slides reviewed by international experts (2.8%, 437/15,697) compared to those that had no international reinterpretation (3.6%, 476/13,064; OR = 0.758, OR 95% CI: 0.663–0.867).
CIN1, CIN2 and CIN3+ prevalence in HR-HPV-positive women
Overall, 4,500 rural women and 357 urban women were positive for HR-HPV (Table 2). Within the subset of rural women, overall crude prevalence was 16.3% for CIN1, 8.7% for CIN2 and 9.8% for CIN3+, whereas the age-standardized prevalence was 19.1% (95% CI: 18.8–19.4%) for CIN1, 8.9% (95% CI: 8.7–9.1%) for CIN2 and 7.6% (95% CI: 7.4–7.9%) for CIN3+ (Table 2). Comparatively, HR-HPV-positive urban women had lower crude and age-standardized CIN1, CIN2 and CIN3+ prevalence (Table 2).
Age-stratified prevalence of CIN1, 2 and 3+ among HR-HPV-positive women was, as before, restricted to the subpopulation of rural women (Fig. 4). CIN1 prevalence among HR-HPV-positive rural women declined with age from 23.8% in 15- to 29-year-old women to 13.0% in 50- to 59-year-old women (Ztrend = −2.713, ptrend = 0.007). CIN2 prevalence was approximately constant with age and ranged between 8.1% in women aged 30–34 years and 50–59 years and 9.6% in women aged 40–44 years (Ztrend = 0.046, ptrend = 0.964). CIN3+ prevalence in HR-HPV-positive rural women steadily increased with age up to 12.8% in 45- to 49-year-old women and then slightly declined (Ztrend = 4.719, p < 0.0001; Fig. 4).
The CIN2+ prevalence among HR-HPV-positive women was slightly higher in studies in which histological slides were reviewed by international experts (18.8%, 420/2,240) compared to those that did not have international experts reinterpret histology (17.6%, 460/2,617), but statistical difference was not shown (OR = 1.073, OR 95% CI: 0.922–1.248).
This pooled analysis is the largest study ever conducted on the prevalence of HR-HPV infection and CIN in China, with 17 study sites in nine Chinese provinces represented. Data were collected in several communities throughout the country, and our study population was composed predominantly of rural women, in agreement with the general female population in the country.5
Compared to the prevalence of HPV (9.9%) in Beijing,22 which is one of the seven cancer registries in China included in Cancer Incidence in Five Continents, the age-standardized HR-HPV prevalence in our study population (16.8%) was much higher, but similar to that in Southern India23 and higher than that in many other low- and intermediate-resource countries in Asia (including Thailand, Nepal, Vietnam and Pakistan).24–26 Only in Mongolia27 was substantially higher HR-HPV prevalence found. When stratified by age and geographic location, age-specific HR-HPV prevalence curves showed a transient decline in rural women aged 25–29 years and 35–39 years in urban women. Bimodal distribution of HPV prevalence is consistent with prior studies from Latin America28 and Asia29 but different from the flat age-specific HPV prevalence curves observed in other studies from Asia.23 Our findings are also at variance with those from recent meta-analyses, which showed that the HR-HPV curve in Asia decreased steadily with increasing women's age.24, 30
The difference in age-specific HR-HPV prevalence by area of living may depend on age-related variations in sexual and reproductive behavior between urban and rural Chinese women. The period of childbearing in China and elsewhere tends be accompanied by stronger family ties and fewer extramarital sexual affairs. Earlier dip in HR-HPV prevalence in rural women may therefore be explained by the fact that age at first intercourse and first birth were 3 and 4 years earlier, respectively, for rural women compared to urban women. Furthermore, more rural women reported having two or more lifetime sexual partners (22%) than urban women (16%). Although our study did not include information on changes husband's sexual behavior after attaining final family size, new husband's extramarital sexual relationships may also contribute to a second rise of HR-HPV prevalence in married women.
In agreement with the relatively elevated prevalence of HR-HPV infection, our study also showed high prevalence of every grade of CIN among Chinese women. Nearly 2% of women in our study harbored CIN3+, an elevated prevalence that is most probably explained by the inadequacy of past and current cervical cancer screening practices in China, particularly among rural women. Of note, the assessment of age-stratified CIN prevalence in our study had to be restricted to rural women not only due to the relatively low number of urban women in individual age groups but also due to comparability problems between studies. In fact, studies in rural areas tended to have more comprehensive histological assessment for the presence of CIN than did those from urban areas.
The prevalence of CIN among HR-HPV-positive women is of special interest as these lesions may eventually progress to cervical cancer. Age-stratified CIN1 prevalence among HR-HPV-positive rural women steadily declined until age 40–44 years, indicating that CIN1 is an early correlate of virus acquisition, particularly among younger women.31 Conversely, similar to previous research,32 in our inadequately screened population, the prevalence of CIN3+ increased with age up to until age 45–49 years. In addition to the fact that the lack of lesion removal through screening may contribute to this pattern, a biological interpretation for the increasing CIN3+ prevalence is the natural effect of aging. More specifically, recent research33 has shown a strong menopause-related effect on cervical cancer risk, with an increasing risk of cancer after incident infection but decreasing overall prevalence of new CIN3+ because CIN3 cases (though not cervical cancer) arise less frequently among older women compared to younger women. As expected, the prevalence of CIN2 among HR-HPV-positive women shows an age-related pattern intermediate between those of CIN1 and CIN3+, supporting prior evidence of the heterogeneity of CIN2 and the concomitant difficulty of distinguishing CIN2 lesions from lower or higher grades of CIN.34
Combined prevalence of CIN2 and CIN3+ in Chinese women (∼3%) was higher than the corresponding prevalence in other Asian countries, where it ranged between 0.5% in Thailand25 and 1.6% in Mongolia.27 In addition, the maximal prevalence of combined CIN2 and CIN3+ was observed in Chinese women older than 40–44 years, which was later than the maximal prevalence of CIN2+ in a recent meta-analysis of Asian women31 and a previous study in Hong Kong.29
The high prevalence of CIN2+ in middle-age Chinese women should inform cervical cancer screening policies in China and other low- and intermediate-resource countries. Current cervical cancer screening recommendations are that women between the ages of 35 and 40 years be screened at least once with cervical cytology, HPV DNA testing or VIA (in low-resource settings).35 However, our findings show that Chinese women aged 40 years or older should also be screened as they harbor a substantial proportion of precancerous cervical lesions that accumulated over time and may still be the target of life-saving treatments. Indeed, a recent population-based study from rural Nigeria showed that the positive predictive value of an HR-HPV-positive test was greater in women aged 50 years or older than in women younger than age 30 and not significantly lower than in 30- to 49-year-old women.36
About 40% of women with undiagnosed CIN3 have been shown to develop cervical cancer.37 Our present findings indirectly raise, therefore, the possibility that cervical cancer incidence in China may be higher than currently estimated (9.6 per 100,000 women1). Chinese cancer registries on which nationwide cancer incidence is estimated include only 2% of Chinese population and are located in five urban settings.1, 3 Residents of these five cities have annual incomes that range from 1.3 to 3 times higher than the national Chinese average,4, 5 suggesting that the population composition of Chinese cancer registries may not be representative of the whole country. Furthermore, some comprehensive data on cervical cancer mortality4 showed that in rural China cervical cancer mortality rates were higher than those for any other cancer in women.38 Although our present study population is more representative of the Chinese population than are the national cancer registry areas, and we carefully evaluated a majority of participating women for HR-HPV and CIN prevalence, residual underestimation CIN3+ prevalence is likely to be present also in our study for different reasons. First, a not negligible fraction of HR-HPV-positive women who did not receive a cervical biopsy were excluded from our present analysis. Second, underascertainment of CIN2+ may have occurred in our study sites (mainly urban sites) where four-quadrant biopsies were not performed.39
Our present pooled analysis has several strengths. It includes 17 population-based studies all conducted by CICAMS that used similarly trained staff and similar screening protocols. Moreover, cytology results and histology results in all studies were verified by senior cytopathologists at CICAMS and/or international experts. Nearly half of final diagnoses derived from biopsy results and four-quadrant punch biopsies were used in selected sites when no suspicious lesion was observed in colposcopy.18 Finally, reliance on three negative screening tests to classify a woman as negative should have allowed good sensitivity level.40 Because the CIN prevalence in our study was relatively higher than that of other Asian countries, we evaluated the potential effect of international reviewing of histology to verify whether overdiagnosis of CIN2+ from Chinese pathologists existed. This analysis showed that CIN2+ prevalence was lower in studies that had international review (2.8%, 437/15,697) compared to those that had not (3.6%, 476/13,064; OR = 0.758, OR 95% CI: 0.663–0.867). However, once the analysis was limited to HPV-positive women, no statistical difference was shown between CIN2+ prevalence in internationally reviewed studies (18.8%, 420/2,240) and that of noninternationally reviewed studies (17.6%, 460/2,617; OR = 1.073, OR 95% CI: 0.922–1.248). Considering that 96.4% of CIN2+ lesions (880/913) occurred in HPV-positive women, the statistical difference revealed by logistic regression modeling in CIN2+ prevalence of all women was probably secondary to the large sample size. Moreover, ten of 17 studies had been reviewed by international experts, so the overdiagnosis was very limited. A limitation to our study is the relatively small size of urban women that limited our possibility to compare CIN prevalence by area of living in individual age groups.
In conclusion, our findings on high prevalence of HR-HPV infection and CIN2+ suggest that the cervical cancer burden in China is heavy, and comprehensive screening and HPV immunization efforts are warranted.
Y.L.Q. and F.H.Z. had full access to all the data in the study and take responsibility for the integrity of the data and accuracy of the data analysis. Our work was supported by the Fogarty International Center at the National Institutes of Health through the Fogarty International Clinical Research Fellows Program at Vanderbilt University (A.K.L. and S.Y.H.); the Academic Capacity Development Program of Beijing Municipal Commission of Education Grant (F.H.Z.) and MOH Special Research Grant (Y.L.Q. and F.H.Z). The authors thank the local doctors and the women who participated in our study. J.L.B. has received support in kind (reagents, testing and funds for direct support and research), under the auspices of Preventive Oncology International, from Hologic, Qiagen, Gen-Probe, Merck and BGI Shenzhen. J.S.S has received grants and consultancy fees or honorarium from Gen-Probe, Hologic, GSK, Qiagen and Merk in the last four years.