Objective Human papillomavirus (HPV) infection causes multiple cancers in both women and men. In China, both HPV vaccination and cervical cancer screening coverages are low. We aim to investigate the temporal and geographical trends of HPV DNA prevalence in heterosexual men, women, men who have sex with men (MSM) and people living with HIV (PLHIV) in China.
Methods We conducted a systematic review, collecting publications in PubMed, Web of Science, China National Knowledge Infrastructure (CNKI) and Wanfang Data from January 2000 to May 2017. A total of 247 studies were selected for this meta-analysis to estimate pooled HPV prevalence, incidence of cervical cancer and risk of infection for subgroups. Meta-regression was applied to identify contributing factors to prevalence heterogeneities.
Results The national HPV prevalence was 15.6% (95% CI (14.4% to 16.9%)) in women with normal cervical cytology, and Central China had the highest prevalence (20.5% (15.2% to 25.8%)). HPV prevalence in heterosexual men (14.5% (11.3% to 17.7%)) was comparable with that of women (OR=1.09 (0.98 to 1.17)), but HPV prevalence in MSM (59.9% (52.2% to 67.6%)) was significantly higher than that in heterosexual men (OR=8.81 (8.01 to 9.69)). HIV-positive women (45.0% (38.4% to 51.6%)) and HIV-positive MSM (87.5% (82.3% to 90.9%)) had 4.67 (3.61 to 6.03) and 6.46 (5.20 to 8.02) times higher risk of HPV infection than their HIV negative counterparts.
Conclusion HPV infection is prevalent in China, particularly in Central China, in comparison with the global level and neighbouring countries. Targeted HPV vaccination for women, MSM and PLHIV and scale-up of cervical screening for women are priorities in curbing the HPV epidemic in China.
- gay men
Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
An estimated 12% of women and 10% of heterosexual men are infected with genital human papillomavirus (HPV) globally.1 2 Approximately 14.1 million new cancer cases were diagnosed in 2012 worldwide, and at least 528 000 (3.7%) were attributed to HPV infection.3 Oncogenic HPV genotypes are responsible for 99.7% of cervical cancer cases, more than 80% of anal cancer cases and approximately 50% of penile cancer cases.4 The incidence of cervical cancer ranges from 3/100 000 to 83/100 000 person-years in China depending on geographical regions.5 Worldwide, the anal cancer incidence is approximately 5/100 000 person-years in HIV-negative men who have sex with men (MSM) and 45.9/100 000 person-years in HIV-positive MSM.6
Previous meta-analyses from 2007 to 2013 demonstrated that the overall HPV DNA prevalence among Chinese women ranged from 13.1% to 18.8%.7–14 There has been no analysis describing the geographical distribution of HPV in China and its burden in heterosexual men.15–18 MSM are disproportionately affected by HPV: an estimated 59.2%19 of Chinese MSM were found HPV DNA-positive in the anal canal. Further, the HPV epidemic overlaps with many other STIs, especially HIV. The host immune response of HPV-positive people living with HIV (PLHIV) is weakened due to the depletion of CD4 T cells, and this may significantly increase the risk of both AIDS and cancers in coinfected individuals.20 To date, very few studies have discussed HIV and HPV coinfection in Chinese high-risk populations, and even fewer reports on the general population.
In 2009, China implemented a state-funded National Cervical Cancer Screening Program in Rural Area (NACCSPRA) for female residents aged 35–59 in 221 pilot counties in all 31 Chinese provinces and autonomous regions,21 but the programme was severely limited by the number of rural surveillance sites, shortage of qualified healthcare personnel and insufficient compensation from the government.22 Not until late 2016 did the China government approve its first commercial bivalent HPV vaccine against HPV16 and 18 for women up to age 26 years in the mainland region. Quadrivalent and nonavalent HPV vaccine has come into the Chinese market in 2017 and 2018.23
This study aims to provide an update of prevaccination HPV prevalence in China, across various geographical locations and populations. Our estimation for the pooled HPV prevalence was based on a systematic literature search on heterosexual men, women, MSM and PLHIV. Previous meta-analyses reported overall HPV DNA prevalence in women, but provided limited information on its geographical distribution, especially the distribution of HPV genotypes. Also, previous studies largely focused on HPV infection in women, not specifying its distribution in high-risk populations of PLHIV and MSM. In this meta-analysis, we aimed to estimate the temporal and geographical trends of HPV prevalence in heterosexual men, women, MSM and PLHIV in China, and compare genotype compositions in women at different cervical precancerous stages. This evidence will inform health policies for HPV screening and vaccination.
This systematic literature review was conducted in peer-reviewed research articles published from January 2000 to May 2017 in PubMed, Web of Science, China National Knowledge Infrastructure (CNKI) and Wanfang Data. We used the following keywords in the search strategy: (‘women’ OR ‘female’) AND (‘CIN’ OR ‘cervical lesion’ OR ‘cervical cancer’ OR ‘cervical intraepithelial neoplasia’), (‘men’ OR ‘male’), (‘MSM’ OR ‘homosexual*’ OR ‘men who have sex with men’ OR ‘gay’ OR ‘transgender’ OR ‘transsexual’), AND (‘HIV’ OR ‘human immunodeficiency virus’). This was combined with additional keywords—(‘HPV’ and ‘human papillomavirus’) AND (‘China’) AND (‘prevalence’ OR ‘incidence’ OR ‘distribution’ OR ‘genotype*’)—to identify the articles reporting HPV prevalence in mainland China. These keywords were translated into Chinese terms accordingly when searching the Chinese databases of CNKI and Wanfang Data. Further, we replaced the keyword (‘China’) with the name of each Chinese province when searching for Chinese literature. An additional 11 records were downloaded from a separate database called CQVIP due to incomplete abstract information in CNKI and Wanfang Data. This systematic review was performed in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement issued in 2009 (online supplementary appendix).
Supplementary file 1
Review articles and non-peer-reviewed local/government reports were excluded when collating records on EndNote X8. We compared the titles and abstracts of records and removed duplicates and records with no abstracts. If data of the same population were reported more than once, the earlier publication was retained. We only selected studies sampling at the cervix for women, and genital and anus for men (including MSM). Testing methods for HPV included both DNA amplification screening and genotyping confirmation. Studies using oral saliva samples and blood antibody testing were not included. We excluded studies that did not report detection methods, study period, study location, having sample size less than 50, and study participants experiencing other infections or diseases such as oropharyngeal, oesophageal and penile cancers. Although we intended to include anal HPV prevalence as one of our outcomes, we did not find any studies specifically reporting anal cancer cases in heterosexual men and MSM.
Data extraction and classification
More than one data point may be extracted from a single study if the study reported information on multiple populations, years, HIV infection status or cervical intraepithelial neoplasia (CIN) clinical stages (figure 1). Information extracted from the articles included (1) name of the first author, (2) publication year, (3) study period, (4) study location, (5) recruitment methods, (6) sampling methods, (7) study type, (8) confirmation detection methods for HPV, (9) age range, (10) residence, (11) HPV DNA prevalence for single type, multiple types, high-risk types, low-risk types and any reported individual genotypes, and (12) cervical clinical stages for women from each eligible article. Genotype confirmatory methods were categorised into seven groups, including membrane hybridisation with amplicon, Hybrid Capture II (HC-II), quantitative real time PCT (RTPCR) sequencing, microarray, enzyme-linked immunology array, multiple detection methods and others. Extracted data points were assigned to a temporal line according to their end study year. China initiated the NACCSPRA from 2009. To assess the impact of this screening intervention, data points were split to pre-2010 group and post-2010 group.
Oncogenic HPV subtypes, including 16, 18, 31, 33, 35, 39, 45, 51, 52, 56, 58, 59, 68, 73 and 82, were defined as high-risk HPV, while the other genotypes were low-risk HPV.24 None of the studies in women reported the sexual orientation of women. Men with no specific sexual orientation reported were regarded as heterosexual. MSM were categorised according to their self-reported sexual orientation in the publications. We extracted data from both HIV-negative and HIV-positive individuals. There were not sufficient data reporting HPV subtypes in HIV-negative women, and we used genotype prevalence in heterosexual women for comparison with HIV-positive women. Studies in heterosexual women were first sorted by provinces, then grouped into the seven geographical regions (East, Northeast, North, Northwest, Southwest, South and Central China). We adopted the cytological examination standard of classification in the 2001 Consensus Guidelines for the Management of Women with Cervical Cytological Abnormalities.25 Precancerous lesions were classified into four stages: NILM (negative for intraepithelial lesion or malignancy), ASC-US (atypical squamous cells of undetermined significance), LG (low-grade squamous intraepithelial neoplasia) and HG (high-grade squamous intraepithelial neoplasia). Female data were also hierarchically classified into four clinical stages according to cytology examination outcomes. We defined participants from population-based or hospital-based studies with NILM as women with normal cytology (NC); women undergoing CIN1 or low-grade squamous intraepithelial lesion as low-grade lesions (LG); women with CIN2/3 or high-grade squamous intraepithelial lesion as high-grade cervical lesions (HG); and women with cervical cancer were abbreviated as CC. We extracted information on the overall HPV prevalence for individual genotype and calculated the multiple (M) and single (S) infection rates. The ratios of multiple infection rate to single infection rate across four cervical clinical stages were defined as ‘MS ratios’. All the HPV prevalences reported in the results refer to viral DNA replication.
Meta-analyses were conducted to synthesise HPV prevalence using STATA/SE V.12.0 for Mac. Heterogeneity of outcomes was tested with the I2 statistics. In all analyses, we selected the random-effects model for the pooled analysis given high heterogeneities in the data. The HPV pooled prevalence and OR with 95% CIs and corresponding p values were recorded. All comparisons between HPV prevalences were assessed with crude OR without adjustment. Whenever data were available, we synthesised the overall HPV and each subtype DNA prevalence by the following states: (1) high-risk and low-risk HPV subtypes; (2) sexual orientation: heterosexual men, women and MSM; (3) HIV status; (4) cervical clinical stages in women (figures 3 and 4); (5) single versus coinfected HPV status; and (6) defined age intervals in women. Meta-regression was used to identify factors that contributed to HPV prevalence heterogeneities in each clinical stage. The variables included the sample size (≤300, >300), study language (Chinese, English), end study years (≤2009, >2010), HPV detection methods and geographical regions (online supplementary table S6). A factor was considered to be statistically significant if the p value was less than 0.05. Publication bias was measured with Egger linear regression.
Our search strategy resulted in 2702 records, with an additional 11 manually selected records. In total, 569 full-text articles were downloaded (n=519 records removed for replicated titles; n=1012 records removed for irrelevant topics; n=613 records removed for absence of key indicators after abstract screening). We finally included 274 eligible full-text articles (256 for women, 4 for heterosexual men and 14 for MSM) (figure 1). Meta-analysis of the national distribution of HPV prevalence was derived from 6706 heterosexual men, 878 169 women (828 003 with NC, 16 668 with low-grade cervical lesions, 20 192 with high-grade cervical lesions, 13 756 with CC) and 4385 MSM. HPV infection status was also reported in 1465 women and 882 MSM living with HIV. Individual study information is included in online supplementary tables S1–S5.
Geographical and temporal trends of HPV infection
Two-hundred and eighty-five data points from 256 selected studies were extracted to demonstrate the geographical distribution and temporal trends of HPV prevalence in Chinese women. Both linear regression and meta-regression showed there were no temporal trends in HPV DNA prevalence in women with NC, LG, HG and CC (online supplementary table S6).
The overall national HPV DNA prevalence in women with NC was 15.6% (14.4% to 16.9%; figure 2). The prevalence was highest in Central China (20.5% (15.2% to 25.8%)), followed by East China (16.9% (14.9% to 19.0%)) and Northwest China (16.5% (12.9% to 20.1%)). Compared with South China, the HPV DNA prevalence in Central and East China was significantly higher (OR=1.09 (1.02 to 1.17), p=0.012; OR=1.06 (1.02 to 1.10), p=0.004; online supplementary table S6). The major high-risk HPV subtypes 16, 18, 52 and 58 showed uneven distributions across the country, although the differences were not significant by meta-regression (figure 2). HPV16 was shown to be significantly more prevalent in Northwest China than the rest of the country (OR=1.03 (1.01 to 1.05), p=0.008; figure 2B). There were insufficient data to inform the temporal and geographical trends of HPV in heterosexual men and MSM.
HPV prevalence by sexual orientation
The HPV DNA prevalence in heterosexual men (14.5% (11.3% to 17.7%)) was comparable with that of women (OR=1.09 (0.98 to 1.17); figure 3A). HPV16, 31, 33 and 52 were more frequently detected in women than in heterosexual men. Other high-risk oncogenic genotypes, including HPV18 and 45 and low-risk types 6 and 11, showed similar prevalence in both populations (figure 3A). In particular, HPV58 prevalence was significantly lower in women (2.1% (1.8% to 2.3%)) than in heterosexual men (3.5% (2.7% to 4.3%), OR=0.59 (0.47 to 0.75); figure 3A).
The HPV DNA prevalence in MSM (59.9% (52.2% to 67.6%)) was far higher than that in heterosexual men (OR=8.81 (8.01 to 9.69); figure 2A). MSM experienced disproportionately high infection rates in almost all reported genotypes, especially in low-risk genotypes of HPV6 (OR=15.83 (11.93 to 21.02)) and HPV11 (OR=16.74 (12.03 to 23.28)). But the prevalence of low-risk genotypes other than HPV6 and 11 was similar between heterosexual men (0.4% (0.2% to 0.6%)) and MSM (0.6% (0.5% to 0.8%)) (OR=1.55 (0.78 to 3.07); figure 4B).
High HPV prevalence in HIV-positive individuals
The overall HPV DNA prevalence in HIV-positive women was about threefold higher (45.0% (38.4% to 51.6%); OR=4.67 (3.61 to 6.03); figure 4A) in comparison with HIV-negative women (15.0% (10.9% to 19.1%)). HPV11 and HPV45 showed similar prevalence in HIV-positive and HIV-negative women, while other genotypes all had considerably higher prevalence in HIV-positive women (figure 4A). Notably, we found significant CIN development among HIV-positive women. The ORs of CIN1, 2, 3 and cervical cancer were 3.03 (2.44 to 3.76), 3.80 (2.68 to 5.39), 3.21 (2.10 to 4.91) and 4.85 (2.80 to 8.41) in HIV-positive women compared with HIV-negative women (figure 4A).
HIV-positive MSM (87.5% (82.3% to 90.9%)) had the highest DNA prevalence of HPV among all reported populations, much higher than that in HIV-negative MSM (51.9% (41.2% to 61.7%); OR=6.46 (5.20 to 8.02)), and the prevalence of all individual HPV genotypes in HIV-positive MSM was significantly higher than their HIV-negative counterparts (figure 4B).
Only one article with 41 cases reported HPV prevalence in HIV-positive heterosexual men, which was insufficient to be included in the meta-analysis.
Changes in HPV distribution throughout cervical clinical stages
Women with abnormal cervical lesions and cervical cancer had significantly higher HPV DNA prevalence than women with normal cervical cytology (15.6% (14.4% to 16.9%) for NC, 69.8% (61.9% to 77.7%) for LG, 86.0% (84.2% to 97.8%) for HG and 88.7% (86.7% to 90.6%) for CC). However, meta-regression did not reveal significant DNA prevalence differences between HG and CC (OR=1.03 (0.99 to 1.07), p=0.177). Notably, the proportion of HPV16 infection increased from 24.5% in NC to 62.6% in women with CC (online supplementary figure S4). HPV16 was also the only genotype with MS ratio consistently lower than 1:1 in all cytological stages (MS ratio of HPV16=0.33, 0.81, 0.61 and 0.42 in NC, LG, HG and CC, respectively; online supplementary figure S6).
Heterogeneities and publication biases
Cervical clinical stage was a strong predictor of HPV prevalence heterogeneities in women. Meta-regressions were conducted for women for four clinical stages separately. In women with NC, geographical region (Central, OR=1.09 (1.02 to 1.17), p=0.012; East, OR=1.06 (1.02 to 1.10), p=0.004; Southwest=1.07 (1.01 to 1.13), p=0.013) and sample size (OR=1.07 (1.02 to 1.12), p=0.007) were significant contributing factors to HPV prevalence heterogeneities (online supplementary table S6). We identified publication biases with significance only in the meta-analysis for women with NC (p=0.001), HIV-positive MSM (p=0.010) and women with high-grade lesions (p<0.001).
This work provides up-to-date evidence to inform potential HPV vaccination and cancer screening programmes for at-risk populations in China. First, over the past decade, the HPV DNA prevalence among women stabilised at a high level nationally, demonstrating a relatively undisturbed epidemic in a setting with limited interventions. Geographically, Central China is most severely impacted by the epidemic. Second, the HPV DNA prevalence in women is comparable with that in heterosexual men, but MSM have a ninefold higher risk of HPV infection than heterosexual men. Notably, HPV16 is the most common oncogenic genotype in women and MSM, in contrast to HPV58 in heterosexual men. Third, HIV-positive women and HIV-positive MSM demonstrate much higher HPV DNA prevalence, compared with their HIV-negative counterparts, with higher potential for HPV-related cancer progression as well.
We reported the national HPV prevalence to be 15.6% (14.4% to 16.9%) in Chinese women with NC. This result is consistent with previous systematic reviews (13.1%–18.8%8–14; online supplementary figure S1) and a large multicentre cross-sectional population study (16.1%8) of Chinese women. The HPV DNA genotype confirmation tests have different sensitivities.26–28 We included genotype confirmatory method as a variable in meta-regression but did not detect significant variations in HPV prevalence across the seven groups. This indicated that genotype confirmatory methods may not be a key contributing factor to the variations in prevalence, whereas other parameters may play a more dominant role. HPV prevalence in Chinese women is substantially higher than the level globally (9.2%–12.0%), in Asia (8.0%–14.0%), and in high-income regions such as Western Europe (9.0%) and North America (4.7%).2 29 30 The 3-year cervical screening coverage in the UK and the USA exceeded 50% by 2017,31 32 and both had initiated HPV vaccine programmes for school girls since 2006 and 2008.33 34 Cervical cancer screening is only accessible for women with medical insurance in urban China, and over 80% of rural women cannot access affordable cervical cancer screening.22 Commercial quadrivalent vaccine licensed for women came into market in 2017 in China. The consistently high level of HPV prevalence over the 2000–2017 period is a direct reflection of the lack of effective HPV preventions.
Consistent with findings of Li et al,14 Central China has the highest burden of HPV in women. In contrast, women in South China may have better access to HPV vaccination through the neighbouring special administrative regions of Hong Kong and Macau, where commercial HPV vaccines have been available for a decades. Similarly, countries sharing borders with China, such as Thailand (6.0%), Korea (8.5%) and Vietnam (9.5%), which have licensed the commercial use of bivalent or quadrivalent HPV vaccine, reported much lower overall HPV DNA prevalence in women.35 In contrast, India has yet to have the HPV vaccine and reported a similar national HPV prevalence (14%) in women as in China.36–42 All this evidence suggests that accessibility to vaccines is a critical underlying factor for the geographical differences in HPV prevalence, but further investigation is required to confirm this. Our comparison indicated consistently high HPV DNA prevalence in women and heterosexual men. In a setting where HPV vaccination for women is high, the heterosexual transmission of HPV is effectively blocked and heterosexual men are also protected by herd immunity even if they did not get vaccinated.43 44 In contrast, HPV transmission in MSM is ‘self-sustaining’ and requires target vaccination.
We reported similar HPV infection risk in women and heterosexual men in China. However, oncogenic HPV types 16, 31, 33 and 52 were prevalent in women; non-oncogenic type 6 was more common in men. In community-based or couples-targeted studies for HPV,45 46 comparisons revealed that cervical HPV infection in women is more prevalent than penile infection in men. Although the penis may be subjected to a higher risk of HPV infection, these HPV genotypes, especially oncogenic types, are cleared more rapidly than cervical infections.47–49 MSM having an almost ninefold higher risk of HPV infection is alarming. MSM in our included studies were young (age 20–40) and approximately 17.8% of them were HIV-infected. Due to the practice of anal sex, MSM are more likely to be infected with HPV at the anus than heterosexual men, and persistent HPV infection at the anus may be a potential reservoir for transmission.50 High rates of coinfection with HIV and other STIs,51 52 less consistent condom use and multiple sex partners16 are all contributing factors to high HPV acquisition rate. Regular STI screening and targeted HPV vaccination for young MSM may be effective measures for HPV prevention.53–55
Our study confirms that HIV-infected women and MSM are at a much higher risk for HPV infection and cancer progression, consistent with previous reports.6 19 56 57 In the Chinese setting, we recommend three actionable measures. First, early initiation of antiretroviral therapy for HIV and improvement of treatment adherence will be effective in reversing the depletion of CD4 T cells, and decrease the likelihood of HPV infection and prevent HPV-related anal and cervical cancers in the long term.58 Second, HPV vaccination has already been shown to be immunogenic and cost-effective in women,59 60 PLHIV and MSM61–63 in many countries. Universal HPV vaccination for school girls and targeted vaccination for high-risk populations should be promoted by the Chinese government. Third, cervical cancer screening in women, including HPV DNA detection as a secondary preventive strategy, should be scaled up for HIV-positive women.55 64–66
This study has several limitations. First, reports on HPV prevalence were noticeably less in inland regions of China and were absent in seven provinces. This may bias the overall estimation of HPV prevalence. Second, the published survey data to date were insufficient to demonstrate the geographical distribution of HPV for heterosexual men and MSM. Third, indepth investigations on the HPV and HIV coinfection epidemics were scarce, and this meta-analysis only partially revealed the coinfection status for high-risk populations in China. HPV screening coverage may be related to medical insurance coverage across China. Fourth, due to limited data availability, we cannot identify the association between the burden of HPV and cervical cancer screening.
HPV infection is prevalent in China, particularly in Central China, in comparison with the global level and neighbouring countries. MSM are shown to be at much higher risk of HPV infection than the heterosexual population, and HIV significantly increases their risk of HPV acquisition. Targeted HPV vaccination for women, MSM and PLHIV as well as scale-up of cervical screening for women are priorities in curbing the HPV epidemic in China.
The human papillomavirus (HPV) prevalence was 15.6% in women with normal cervical cytology across China; HPV prevalence is highest in Central China (20.5%).
HPV prevalence in men who have sex with men (MSM) (59.9%) was nine times higher than that in heterosexual men (14.5%).
HIV-positive women (45.0%) and HIV-positive MSM (87.5%) are at 4.67 and 6.46 times higher risk for HPV infection than their HIV-negative counterparts.
Targeted HPV vaccination for women, MSM and HIV-positive individuals and scale-up of cervical screening for women should be priorities in China.
XM and QW contributed equally.
Handling editor Jackie A Cassell
Contributors XM and QW were both liable for the data collection, data cleaning, statistical analysis and figure plotting. PP reviewed and double-checked the literature search strategy. XM and LZ wrote the first version of the manuscript. JJO, CKF, SS, PP, LZ and NNS reviewed the manuscript and contributed to modification. JJ, LW and FC reviewed and commented on the draft. LZ and FC cosupervised the study.
Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.
Competing interests None declared.
Provenance and peer review Not commissioned; externally peer reviewed.