The burden of cervical cancer in China has not been characterized in detail. We reviewed cervical cancer data from national mortality surveys and registries, and conducted a meta-analysis to estimate the prevalence of high-grade lesions (HSIL) and high-risk human papillomavirus (HR-HPV) infections in rural Shanxi Province. We found that a national survey in the 1970s estimated age-standardized cervical cancer mortality rates as ∼15 and ∼83/100,000 women nationally and in Xiangyuan, Shanxi; but the latest survey (2004–2005) found much lower rates of ∼3 and ∼7/100,000, respectively. IARC registries record age-standardized cervical cancer incidence in China as <5/100,000 (1998–2002); but the five registry sites cover <2% of the population, and the gross domestic product per capita at each of the registry sites is higher than China's average (by a factor ranging from 1.3 to 3.9). The pooled estimate of the prevalence of HSIL and HR-HPV in women aged 30–54 years in Shanxi was 3.7%(95%CI:2.7–4.8%) and 17.2%(95%CI:13.1–21.3%), respectively. Based on a feasible range informed by the incidence data for China and other unscreened populations, the predicted indicative annual number of new cervical cancer cases nationally, in the absence of any intervention, ranges from ∼27,000 to 130,000 (2010) to 42,000 to 187,000 (2050). In conclusion, recent data suggest comparatively low rates of cervical cancer incidence in China, which may be partly explained by the location of registry sites in higher socioeconomic status areas. However, the evidence is consistent with considerable heterogeneity within China, with a higher disease burden in some rural areas such as Shanxi. Therefore, the lower reported rates of cervical cancer in China should be interpreted cautiously.
Estimates of the worldwide burden of cervical cancer suggest that ∼529,800 incident cases and 275,100 deaths due to cervical cancer occur annually.1 Given that the People's Republic of China contains approximately one fifth of the world's population, the disease burden in this country will have a substantial effect on global estimates of the current and future burden of the disease. A recent estimate by GLOBOCAN 2008 for cervical cancer case numbers in China, based on the most recent mortality survey information and mortality to incidence ratio data, predicted that the number of new cases in China was 78,400 in 2010, which was predicted to rise to 93,500 in 2030.1
Based on the first Chinese national mortality survey conducted in the 1970s,2, 3 it has been thought that rates of cervical cancer are extremely high in parts of rural China. However, two subsequent national mortality surveys have been performed, and these suggest more modest estimates.4, 5 There is no established national or provincial level organized cervical cancer screening program in China; although some, mainly urban, subgroups have high levels of opportunistic screening. Some women have access to employee-based screening through large corporations or organizations, and recently a government-sponsored large-scale demonstration project covering 10 million rural women for visual inspection and cytology-based screening was implemented.6 Despite these initiatives, the majority of the population do not have access to regular screening.
The aim of this study was to review data sources and to characterize, as far as possible, the extent of the disease burden of cervical cancer in mainland China. Therefore, we reviewed the available epidemiological data on mortality and incidence rates of cervical cancer in rural and urban populations, and we used the incidence rate information to estimate a feasible range for the current and future case numbers of cervical cancer that will be diagnosed in China, taking into account the uncertainties in data sources. Because the majority of the population in China (64%) are resident in rural areas (based on data from the fifth population census, conducted in 20007), we also reviewed information specific to rural China, taking Shanxi Province in central-north China as an example of a rural population. Shanxi is a predominantly rural area with an active mining industrial base. Because it has been thought to harbor high rates of cervical cancer, a number of large-scale clinical studies have been performed there since the 1990s.8–14 Therefore, Shanxi is the area in rural China for which the most high quality information on high grade squamous intraepithelial lesion (HSIL) and human papillomavirus (HPV) prevalence is available. Using the data from Shanxi, we synthesized information on rates of HSIL and HPV infections.
ASRC: age-standardized rate adjusted to the Chinese standard population; ASRW: age-standardized rate adjusted to the world standard population; CHIS: the Center of Health Information and Statistics; CI: confidence interval; CIV-IX or -CIV-VIII: IARC Cancer Incidence in Five Continents (CIV) database (Volume 9 or Volume 8); GDP: gross domestic product; HPV: human papillomavirus; HR-HPV: high-risk HPV; HSIL: high-grade squamous intraepithelial lesion; IARC: International Agency for Research on Cancer; ICD: International Classification of Diseases; NCCR: National Center for Cancer Registries of China; UN: United Nations; WHO: World Health Organization
Material and Methods
Data sources and estimates of cervical cancer mortality in China
We reviewed the main sources of information on cervical cancer mortality in China. Three national retrospective mortality surveys have been conducted since the 1970′s, over the periods 1973–1975,2, 3 1990–19924 and 2004–2005.5 For this study, we reviewed the coverage and sampling methodology, and summarized the cervical cancer mortality rates obtained from the three surveys, focusing on summarizing national rates as well as the available information for Shanxi Province.
There are also two ongoing national mortality surveillance systems in China: the Center of Health Information and Statistics (CHIS) reporting system, which is affiliated to the Ministry of Health of China; and a surveillance channel using the Disease Surveillance Points system coordinated by the Chinese Center for Disease Control and Prevention (China CDC system).15 The CHIS data has been included in the World Health Organization (WHO) mortality database since 1987.16 The CHIS and CDC mortality surveillance systems involve a 1% and 10% sample of the total Chinese population, respectively. A comparison of the CHIS and CDC surveillance systems has nevertheless found that the sex- and region-specific mortality rates obtained from the CHIS data were closer to the second national mortality survey than rates obtained from the CDC system.15
In the current analysis, we used the CHIS (WHO) data to assess trends over time in both rural and urban populations, and calculated the percent changes in the average age-standardized mortality rates of cervical cancer over the decade from 1987–1989 to 1997–1999. We stratified the results in two ways: by the age groups <50 and 50+ years; and by <65 and 65+ years; these age groups were chosen for the analysis because the median age at death from cervical cancer in the third national mortality survey was ∼50 years (2004);5 but the median age was ∼65 years over the period 1987–1999 in the CHIS data.
Data sources and estimates of cervical cancer incidence in China
To summarize the available information on cervical cancer incidence in China, we reviewed the data from the cancer registry sites included in the International Agency for Research on Cancer's (IARC's) Cancer Incidence in Five Continents (CIV) which reported on rates of cervical cancer.17, 18 Registry sites based in Shanghai City, Tianjin City and Qidong County from mainland China were included in Volumes VI-VII; and three more sites (Beijing City, Jiashan County and Wuhan City) were added to Volume VIII (1993–1997).17 In the latest volume, CIV Volume IX (1998–2002), a total of five registries were included (Shanghai, Jiashan, and the newly added Harbin City, Zhongshan City and Guangzhou City sites).18 In this study, we reviewed the available data for the individual sites included in CIV Volumes VIII and IX.17, 18 As a proxy measure of the socioeconomic status of each site, we calculated a ratio of the site regional gross domestic product (GDP) per capita to the China GDP national average, using data extracted from local and central Chinese governmental websites.7
In addition to the cancer registries included in IARC's CIV series, there are a larger number of cancer registry sites (more than 30) reporting into the Chinese National Center for Cancer Registries (NCCR) system, and some data from this larger group of registries has been reported.19–27 We extracted and summarized the information on cervical cancer incidence and mortality from these registries.19–27
We also estimated a feasible range for the number of current and future cervical cancer cases in China, applying United Nations (UN) population projections (from 2008)28 and using appropriate information sources to inform a feasible range of age-specific cervical cancer incidence rates.18 To calculate a lower bound for the feasible range, we applied the average age-specific rates across the five China registry sites in IARC's CIV-IX (the overall age-standardized rate across these five sites was 3.2 per 100,000 women, standardized to the world standard population (ASRW, according to Segi)).18 As an upper bound for the range, and to take into account that the incidence rates in some rural populations and in lower socioeconomic status areas in China may be higher than those reported in the IARC registries, we assumed that the overall national age-specific incidence rates in China were equivalent to the average rates observed in less developed countries as defined by the UN, including 27 countries/regions in Africa, Asia, South America and elsewhere, but excluding China28 (the overall average age-standardized rate across these countries was 15.7 per 100,000 women (ASRW)18). For these case projections we assumed an absence of any major future preventative interventions such as HPV vaccination or organized cervical screening programs; and also assumed that no other factors of major relevance would change substantially over next four decades; these included the age-specific pattern of HPV infection, sexual behavior, smoking behavior and use of oral contraceptives.29, 30
Data sources and estimates of the prevalence of cervical precancerous disease (HSIL) and HPV infection
The prevalence of HSIL should provide some indication of the overall cervical cancer risk in an unscreened population, since it is known that a proportion of HSIL lesions (particularly cervical intraepithelial neoplasia grade 3 or CIN3) will progress to invasive cervical cancer if left untreated.31 Furthermore, although a high proportion of HPV infections naturally regress, and cohort effects mean that younger women may have different HPV exposure than their older counterparts, some level of correlation between high risk HPV (HR-HPV) prevalence and cervical cancer incidence in older women has been reported across different countries and settings (excluding China)32; this was increased in women over age 45 years. Although HR-HPV infection is causally responsible for the development of invasive cervical cancer, this correlation will be weakened in the presence of cervical cancer prevention programs. Nevertheless, in a country such as China in which no national prevention programs are in place, the population prevalence of HR-HPV infection and of detected precancerous disease is potentially informative and to some extent may be indicative of the burden of cervical cancer.
As an example of a rural population, we reviewed data on rates of HSIL and HPV infections in Shanxi Province. For the current analysis, we synthesized the information from previous studies by performing a meta-analysis to estimate the overall prevalence of HR-HPV infection and HSIL in Shanxi Province. We also assessed the available information on HSIL and HR-HPV prevalence in other rural and urban populations in China, and compared the available information from China with data from other less developed countries; for this comparison we focused mainly on the prevalence surveys conducted by the IARC.
HPV infection correlates and cofactors
We reviewed the available information on behavioral correlates of HPV infection in China, and exposure to the HPV co-factors that have been associated with progression to cervical cancer, including parity, tobacco use, use of oral contraceptive, and a younger age at first full-term pregnancy.29, 30 We also assessed the available information on the trends for changing exposure to these risk factors in China.
Estimates of cervical cancer mortality in China
The first national mortality survey was conducted in 1975, and involved almost complete population coverage, encompassing ∼96% of the Chinese population (Table 1). This mortality survey was an enormous undertaking involving the retrospective classification of ∼20 million deaths.2, 3 A large cohort of health workers retrospectively assessed cause of death for all deaths occurring in the period 1973–1975, using a standardized questionnaire. Information was obtained from various sources within the health system and by conducting interviews with family members and community leaders.19 Malignant tumors were classified in one of 20 tumor categories, which included cancer of the cervix uteri but did not include specific categories for any other gynecological cancer (although a general category for “other malignant tumor” was available).19 The level of misclassification of cervical cancer as cause of death in the survey is unknown; but the mortality rate from cervical cancer recorded (14.6 per 100,000 women, age-standardized to the world standard population) was approximately one quarter of the overall cancer mortality rate recorded in females.2, 3 It seems possible that over-attribution of cervical cancer as the cause of death could potentially have occurred via misattribution of other gynecological cancers, although it is also possible that some deaths from advanced cervical cancer could have been classified as other malignant tumors. This survey found cervical cancer mortality rates to be 14.6 nationally and 29.4 across Shanxi Province.2, 3 The mortality rates in the closely adjoining counties of Xiangyuan and Yangcheng in Shanxi were 83.0 and 30.5 per 100,000 women, respectively (Table 1).2, 3
Table 1. Summary of the three national mortality surveys in China and the findings for cervical cancer
The second and third national mortality surveys covered the periods 1990–1992 and 2004–2005, respectively, and these both used a multi-stage stratified cluster sampling method with coverage of roughly 10% of the total Chinese population (Table 1).4, 5 As was the case for the first survey, health workers retrospectively assessed cause of death using a variety of sources, but in the second and third survey, cancers were classified using the International Classification of Diseases system (ICD)-9 (second survey) and ICD-10 (third survey).4, 5 The estimated national rates of cervical cancer mortality in the second and the third surveys were 4.3 and 2.5 per 100,000 women, respectively (ASRW).4, 5 These two surveys sampled different counties within Shanxi Province; the second survey found the mortality rate to be 40.7 per 100,000 women in Yangcheng (broadly similar to the findings of the first survey for that country); but the third survey found the mortality rate in Xiangyuan to be ∼7 per 100,000 women, which was much lower than the rate found for the same county in the first survey.2–5 Because of these differences in the findings, cervical cancer data from the various national surveys should be interpreted cautiously.
The CHIS database also gives information on mortality rates in China. Although more than 60% of the total population of China were resident in rural areas in 2000,7 the size of the sampled population in the CHIS database is approximately the same from rural compared to urban areas. Additionally, site participation is voluntary and tends to involve sites on the Eastern seaboard,15 and therefore the database is not completely representative of the national population. However, despite the limitations, these data provide the only consistent source of information which is available over a number of years; therefore this is the only data source that can be used for assessing trends in mortality. Based on the CHIS data, the age-standardized cervical cancer mortality rates (across all ages) in 1999 were 4.2, 1.8 and 2.7 per 100,000 women for rural areas, urban areas and both combined, respectively (data after the year 2000 are not available).16 We found that in the rural Chinese populations, the average mortality rate from 1987–1989 to 1997–1999 remained stable in women under 50 years (change ∼2%) at less than 2 per 100,000 women (Fig. 1). However, the mortality rate reduced by 34% in women aged over 50 years over the period. In the urban populations, although rates in younger women also remained low and stable over the period (<1 per 100,000 women), the reduction in mortality in women over 50 years of age was greater, at 61% (Fig. 1). Similar trends were observed when stratifying by the age groups <65 years and 65+ years. In general terms, the CHIS mortality data suggests a trend for declining rates of cervical cancer mortality in older women; with possibly a more pronounced effect in urban populations.
Estimates of cervical cancer incidence in China
The available cancer incidence registry data from the IARC registries suggests incidence rates of less than 5 per 100,000 women for sites located in both urban and rural areas (Table 2). The cervical cancer incidence rates (standardized to the world standard population, ASRW) reported in CIV-VIII (1993–1997) ranged from 1.2 (Jiashan County) to 3.9 (Wuhan City) per 100,000 women; and those reported in CIV-IX (1998–2002) ranged from 2.4 (Jiashan County) to 4.6 (Guangzhou City) per 100,000 women, with an average rate reported in CIV-IX of 3.2 per 100,000 women.18 Because only two sites (Shanghai City and Jiashan County) contributed data to both CIV-VIII and CIV-IX, it is difficult to perform comparative analysis of incidence rate or time trends using these data.
Table 2. Summary of the reported cervical cancer incidence for the cancer registries in China included in CIV-VIII and CIV-IX, and the ratio of the GDP per capita at each registry site compared to China's average
Registries reporting on cervical cancer rates are a subset of the total number of cancer registries in CIV, and these cover less than 2% of China's female population; with coverage of 8.9 million women in CIV-VIII and 6.1 million women in CIV-IX, compared to a population of 602.3 million women nationally in 2000.7 The gross domestic product (GDP) per capita of the registry sites included in CI5-VIII and IX were higher than the GDP national average (which was 18,934 Chinese Yuan in 2007);7 the ratio of the GDP per capita for the nine registry sites in 2007 compared to the average for China was found to range from 1.3 to 3.9 (Table 2).
Other than the sites selected for inclusion in IARC's CIV, there are a larger number of registry sites in China reporting to the China NCCR. However, even if all of these 30–37 sites are considered (the number of sites reporting varies slightly from year to year), the total population coverage remains less than about 6% of the national population.21 In the China NCCR reports of 1998–200220 and 2003,22, 23 only the incidence rates of the ten most common cancers were presented; at the majority of sites, cervical cancer incidence ranged from 1.9 to 8.4 per 100,000 women (and mortality ranged from 1.2 to 7.0 per 100,000 women). Higher cervical cancer incidence rates of >10 per 100,000 women were recorded for a few registry sites, including Yangcheng County in Shanxi Province and Shenzhen City in Guangdong Province.20,22,23 In the latest three reports (2004, 2005 and 2006), no site-specific information was presented, but the overall cervical cancer incidence rate ranged from 6.0 to 9.1 per 100,000 women.24–27 No substantial differences between rural and urban populations were noted in these data (for more information on the NCCR reported rates and for the corresponding mortality rates, refer to Appendix Table A1).
To estimate a feasible range for the current and future number of cervical cancer cases in China, UN population predictions were used and two sets of age-specific cervical cancer incidence rates were applied (see Fig. 2a). Under the two sets of age-specific cancer incidence assumptions, the estimated number of cases in 2010 ranged from 26,900 to 130,000 and is projected to reach 42,100 to 186,600 in 2050 (Fig. 2b). These projections were made under the assumption that the population structure was determined by fertility remaining at ∼1.8 births per woman until 2050;28, 35 however, under alternative assumptions about future fertility rates, no substantial changes to future projected case numbers were observed (<3% change in maximum projected case numbers in 2050), because the feasible future fertility rates vary within a somewhat constrained range (∼1.4–∼2.3 births per woman).28, 35 The projection did not take account of the effect of these potential changes in fertility on the risk of cervical cancer. However, because the upper end of the projected fertility rates in the UN projection was 2.3, this cofactor is not expected to have a major effect on the estimates of future cervical cancer cases, because according to the pooled worldwide data on the parity-associated risks of cervical cancer, a relatively high number of full term pregnancies are required before risks are substantially elevated.29
A projection of future cervical cancer case numbers in China (to 2030) has also recently been published as part of the GLOBOCAN 2008 estimates;1 this projection used cervical cancer rates obtained from China's third national mortality survey and rural and urban estimates for mortality to incidence ratios (from 36 Chinese cancer registries, 2003 – 2005; not all of which are IARC-certified) to project cervical cancer incidence.1 The resulting estimated case numbers ranged from 78,400 (in 2010) to 93,500 (in 2030),1 and these findings are also depicted in Figure 2b for comparison. It can be seen from Figure 2b that the GLOBOCAN estimates are in the mid-range of the current projections.
Estimates of the prevalence of cervical precancerous disease (HSIL) and HPV infection
We estimated the prevalence of cytological or histological HSIL and HR-HPV in Shanxi Province (as example of a site in rural China) by using random effects meta-analysis to integrate the findings of a number of prior studies conducted in the province.8–13 We found the pooled summary estimates of HSIL and HR-HPV prevalence in women aged 30–54 years in this setting to be 3.7% (95%CI: 2.7–4.8%) and 17.2% (95%CI: 13.1–21.3%), respectively (Fig. 3). The highest HPV prevalence was reported in a study by Belinson et al., 2003,9 and the lowest prevalence was reported by Moy et al., 2009.13 Excluding these outliers in sensitivity analysis resulted in overall HPV prevalence estimates which were broadly similar to the main findings–4.0% (95% CI: 2.6–5.4) for HSIL and 17.0% (95% CI: 16.1–17.9) for HR-HPV infection.
From the national perspective, a previous study has pooled data from six urban and eight rural sites in China,11 which included three individual IARC studies,36 five START studies,13 and five SPOCCS-III studies.37 Appendix Table A2 shows the published findings for individual studies of HPV infection and HSIL conducted across China. The crude estimate of the overall prevalence of histological HSIL in women aged 17–59 years was reported to be 2.5% (424/17,334).11 We also used the information reported in the national pooled analysis to calculate the crude prevalence of HR-HPV infection (assessed via Hybrid Capture 2 (hc2), QIAGEN, Gaithersburg, USA); this was found to be ∼14.5% in both younger (<45 years) and older age groups (≥45 years).11
We compared the data from Shanxi to information for other regions in China. As shown in Appendix Table A2, there are indications that the rates of HPV infection in some other regions of China13, 36 may be comparable to Shanxi Province, although there appears to be considerable variation between the different rural and urban populations. It should be noted that consistent information on HPV prevalence, HSIL and cancer incidence and mortality is not available for most parts of China; because in most of the regions where the HPV/HSIL prevalence studies were performed, there are no IARC-certified registries. Some local registry information is available in these regions (Appendix Table A2), but data from these local registries should be interpreted cautiously.
We also compared these findings to a number of other studies performed in low resource settings in other countries. A total of eight studies (ten populations) were included in this review – of these seven were conducted by IARC and one was conducted by the US National Cancer Institute (NCI) (Table 3).38–45 The comparison indicates that the estimated prevalence of HSIL in Shanxi Province (3.7%) and in China overall (2.5%) (which were generally conducted in females > 15 years of age), appears higher than in many other settings with unscreened populations, which are all 1.8% or lower in the studies examined here. Similarly, the HR-HPV infection prevalence in Shanxi Province, and in China nationally among women over 45 years of age was higher than for the majority of the other settings (Table 3).
Table 3. Comparison of prevalence of high-risk HPV and HSIL in China and selected lower resource setting populations
HPV infection correlates and cofactors
Several HPV infection correlates and co-factors in HPV progression could influence local rates of cervical cancer. There are indications that the number of lifetime sexual partners has increased in young women in China, especially in urban areas, with changing attitudes since the late 1970′s. The age at first sexual intercourse, another variable potentially changing over time, is an important determinant of HPV exposure; and a recent multi-centre survey conducted in mainland China (N = 11,852) has observed an earlier reported age of sexual debut in younger cohorts (the average age of sexual debut was 16.6, 19.3 and ∼21 years for women aged 15–19, 20–24 and 24–59 years, respectively, surveyed in 2009).46
Multi-parity is one of the confirmed HPV co-factors;29 and data from the UN suggests an average fertility rate for China's women of 1.8–2.0 births per woman in the past two decades, which is lower than the average rate worldwide (2.6–3.1).35 This is potentially a result of the “one-child” family planning policy initiated at the end of the 1970s. The one-child policy has been impacting on urban populations much more than on rural populations, and has potentially contributed to the reduction in the incidence of cervical cancer observed in urban Chinese populations (Fig. 1). The fertility rate remains much higher in rural China (3.1–4.0,35 higher than the global average), which implies that women in rural China may not have experienced substantial reductions in risk of cervical cancer due to reductions in parity.
A younger age at first full term pregnancy has also been found to be a co-factor in HPV progression (independently of parity).29 Although specific information is not available for this exposure, based on UN data during 1995–2000, 2.5% of births in China were delivered by mothers aged between 15 and 19 years28 (2.8% in rural and 0.9% in urban areas, according to Chinese national Census data in 200021). This rate is much lower than the average of 15.2% in less developed regions (excluding China), and is also lower than the average of 8.5% for more developed regions;28 suggesting that women in China are exposed to relatively low risks due to this potential co-factor.
Exposure to other HPV co-factors such as smoking47 and oral contraceptive use10, 12, 14 is also reported to be very low among Chinese women. Overall, reductions in parity, better cervical cancer treatment and some opportunistic screening could potentially have contributed to the mortality decline observed in older urban women in Figure 1, but it is unclear what factors may be involved in the lesser, but still substantial, declines observed in recent decades in older rural women.
The extent of burden of cervical cancer in China has not previously been characterized in detail. This review of the available data sources and projection of future case numbers of cervical cancer is essential to understanding the current and future burden of disease, and provides important background to any assessment of the potential impact of future cervical cancer prevention and control strategies in China, including the possible introduction of cervical screening programs or of HPV vaccination initiatives.
The first national mortality survey was an unprecedented source of information on cancer death in China. However, it is possible that this survey could have overestimated cervical cancer mortality in some regions, due to diagnostic misclassification. In contrast, the second and the third mortality surveys could have potentially underestimated the mortality from cervical cancer. Only 11 sites contributed to all three mortality surveys,48 and therefore it is difficult to compare the survey findings for cervical cancer mortality over time and across surveys. However, the latest mortality rates for China from the third survey are low even compared to well-screened developed country populations. For example, the mortality rate of 2.5 per 100,000 women16 reported in the third mortality survey is well within the range observed in most developed countries—for comparison, in the USA, UK, Australia, Finland and Singapore, age-standardized mortality rates range from 1.0 to 3.7 per 100,000 women5, 16 (Appendix Table A3).
Some specific regions in rural China, particularly Shanxi Province, have been thought to have amongst the highest rates of cervical cancer in the world. However, we have shown that the evidence for this is mainly derived from historic data from the first and second mortality surveys, and from the NCCR registries (not all of which are certified for inclusion in IARC's CIV reports). In this study, we have discussed some methodological issues associated with the collection of these data. We recommend that the early data from the first survey showing extremely high rates of cervical cancer mortality in Shanxi province (>50 per 100,000 women) need to be interpreted cautiously, particularly since it does not appear consistent with the much lower mortality rates recorded in the third survey. Mortality rates for Shanxi Province in the third survey, although relatively high compared to other regions assessed in the survey, were still rather low (<7 per 100,000 women) compared to other unscreened populations outside China.
Similarly, both the IARC cancer registries and the larger number of registry sites reporting to the NCCR record relatively low cervical cancer incidence rates. The latest reported cervical cancer incidence rate for China in IARC's Cancer Incidence in Five Continents (3.2 per 100,000 women18) is lower than rates for most screened populations in developed countries, which are between 4.0 and 10.6 per 100,000 women18 for the USA, UK, Australia, Finland and Singapore (Appendix Table A3). We found that, although the data from IARC registries suggest relatively low cervical cancer incidence rates, that the IARC registries reporting on cervical cancer cover less than 2% of the national population. Furthermore, we found that the five registry sites reporting on cervical cancer are located in areas with a substantially higher GDP per capita compared to China's average. Therefore, issues of population representivity must be considered when interpreting the data; and these issues may partly explain why low cervical cancer incidence rates, even compared to developed country populations, are reported for China.
Despite the fact that the data from recent mortality surveys and registry data record relatively low cancer incidence and mortality rates in both rural and urban areas, we have shown that the available data on surrogate measures of the risk of cervical cancer, such as the prevalence of HR-HPV and HSIL, suggests that the burden of disease may be more substantial than the registry information would suggest. When the rates of HSIL in women aged 15–54 years from six urban and eight rural sites in China are compared to unscreened populations in other developing countries, they are considerably higher in China (2.5% for China and 3.7% for Shanxi vs. <2.0% for the other populations). This is despite the fact that registry data for these other countries indicate much higher cancer incidence rates compared to China, with the majority of the other countries recording cancer incidence rates of between ∼15 and ∼60 per 100,000 women (Table 3).1, 17, 18, 33, 34 If it is assumed that there is a strong correlation between rates of endemic high grade precancerous disease and cervical cancer incidence rates in unscreened populations, and if these data are indicative of such a relationship, then the high rates of HSIL observed in China would suggest cancer incidence rates, at least in some areas, which are much higher than the average of 3.2 per 100,000 women recorded in the latest IARC data. Similarly, our findings show that rates of endemic HPV infection in both rural and urban populations in China were higher than in many other populations. Prior work has suggested a correlation between rates of HR-HPV infection in older women and cervical cancer incidence,32 although it should be noted that the relationship between HR-HPV prevalence and cancer incidence may be heavily confounded and/or affected by various issues including data quality and measurement error for HPV studies;32 as well as sampling strategy, co-factor exposure, cohort effects and any opportunistic screening in the population. Nevertheless, our findings do seem to suggest a higher burden of disease than that reflected in the available registry data.
Overall, we conclude that the data sources on the burden of cervical cancer in China are limited, and recent estimates for national mortality and incidence rates should be interpreted cautiously. Some rural areas in particular, such as Shanxi, appear to have a high burden of HPV infection and precancerous disease, and the most recent information for Shanxi suggests cervical cancer mortality rates that, whilst relatively low compared to unscreened populations outside China, are approximately double the estimated national mortality rate. Recent incidence data for both rural and urban Chinese populations suggest relatively low cervical cancer rates compared to highly screened populations in developed countries, which may be partly explained by the location of cancer registries in areas with high socioeconomic status populations. Because of these uncertainties, the predicted current and future burden of cancer in China spans a wide feasible range.
Given the importance of the contribution of China to the worldwide estimates of the burden of cervical cancer, in the future it will be very important to improve the cancer surveillance system and cancer registry systems in China. Consideration could be given to focusing on quality certifications and IARC registrations for a larger number of local registries, particularly those that that are operating in lower socioeconomic status areas (with a lower GDP per capita). This would have the effect of expanding certified registry coverage nationally and would result in better population representivity of the information reported to IARC/WHO. This information will be of paramount important in assisting policy decision-making in relation to future cervical cancer prevention initiatives in China.
Dr. Shi was funded by a UICC American Cancer Society Beginning Investigators Fellowship funded by the American Cancer Society (ACS/09/008) and by Cancer Council NSW, Australia.
Table A1. Summary of the National Center for Cancer Registries of China (NCCR) data on cervical cancer rates
Table A2. Measures of prevalence of high-risk HPV infection and HSIL, cervical cancer incidence and mortality, by location in China1
Table A3. Comparison of reported cervical cancer mortality and incidence rates for China and selected populations, per 100,000 women