Geographic and temporal variations in cancer incidence attributable to human papillomavirus infection in China: a population-based study Authors: Junli Liu, Junrong Hu, Qixun Zhu, Runquan Chen, Junyao Ye, Yihuan Zheng, Feixue Wei, Ganfeng Luo, Yanting Zhang Published: 2026 ABSTRACT BackgroundCurrent epidemiological data on cancer incidence attributable to human papillomavirus (HPV) infection in China remain unclear. This study provides an updated assessment of the magnitude of and geographic variability in the incidence of HPV-attributable cancer in China in 2013–2017, temporal trends from 1988 to 2017 and projections to 2032.MethodsData were extracted from the Cancer Incidence in Five Continents Volumes VII–XII. The population-attributable fraction was calculated by multiplying the high-risk HPV prevalence at each anatomical site by the specific biomarker positivity rate in HPV-positive cases. Joinpoint regression and age-period-cohort models were used to quantify temporal age-standardized incidence rate (ASR) trends and age and birth cohort effects, respectively. Bayesian age-period-cohort models were constructed to project the HPV-attributable cancer incidence through 2032.ResultsAn estimated 68,955 and 3463 HPV-attributable cancer cases were observed among females and males in China in 2013–2017, respectively, with ASRs of 11.20 (95% confidence interval [CI]: 10.98–11.41) and 0.52 (0.20–0.82) per 100,000. The most common HPV-attributable cancers were cervical cancer (females; ASR [95% CI]: 10.92 [10.83–11.00] per 100,000) and laryngeal cancer (males; 0.21 [0.03–0.37] per 100,000). Among the 25 provinces included, the ASRs of HPV-attributable cervical cancer ranged from 6.15 (5.88–6.42) per 100,000 in Beijing to 17.43 (15.84–19.03) per 100,000 in Hunan (2.8-fold difference). The provincial Human Development Index was inversely associated with the cervical cancer incidence in multivariable linear ecological regression analyses. From 1988 to 2017, the ASRs increased for HPV-attributable cervical cancer (females; average annual percent change [AAPC] and 95% CI: 5.89% [2.82–9.06]) and penile and oropharyngeal cancers (males; 1.98% [0.82–3.15] and 9.65% [5.81–13.62], respectively). Substantial increasing risks across successive generations born between 1908 and 1978 were observed for HPV-attributable cervical cancer among females and oropharyngeal cancer in both sexes. Under a counterfactual continuation of historical trends through 2017, without incorporating subsequent vaccination or screening expansion, the ASR of HPV-attributable cervical cancer is projected to reach 19.53 (95% uncertainty interval [UI]: 9.99–34.36) per 100,000 in 2028–2032, compared with 10.92 (95% CI: 10.83–11.00) per 100,000 in 2013–2017.ConclusionsIn China, the incidence of HPV-attributable cancer varies considerably according to cancer type and geography. These findings highlight the need for region-specific HPV vaccination and cancer screening strategies, particularly in areas with high incidence rates.Supplementary InformationThe online version contains supplementary material available at https://doi.org/10.1186/s12889-026-29368-9. Introduction Human papillomavirus (HPV) infection is a major cause of cancer in both sexes. In 2018, HPV infection was responsible for an estimated 694,000 new cancer cases worldwide, for an age-standardized incidence rate (ASR) of 8.0 per 100,000 person-years [1]. In China, the number of HPV-attributable cancer cases was 110,000 (ASR: 5.6 per 100,000) [1]. This substantial disease burden underscores the importance of research on HPV-attributable cancers in China according to geography and cancer type. HPV is a well-established epidermal agent of multiple benign and malignant diseases that primarily infects the squamous epithelium [2]. Among the known oncogenic HPV genotypes, HPV16 accounts for the majority of HPV-attributable cancers worldwide [3]. A series of prevention strategies have been implemented in China, including the launch of the Cervical Cancer Screening Programme in Rural Areas in 2009 and the approval of bivalent, quadrivalent, and 9-valent HPV vaccines since 2016 [4, 5]. The inclusion of HPV vaccination in the national immunization program (NIP) on November 10, 2025, has further accelerated progress towards cervical cancer elimination [6]. Previous studies have advanced the understanding of the HPV-attributable cancer burden in China. Lu et al. quantified the national burden of HPV-attributable cancers during the 2005–2015 period, providing important evidence on the overall numbers and temporal trends of HPV-attributable cancers in China [7]. A recent multinational analysis by Tian et al. characterized long-term incidence trends of HPV-attributable cancers across 44 countries (including China) via the Cancer Incidence in Five Continents (CI5) Plus database [8]. However, existing studies have focused mainly on national-level estimates or cross-country comparisons; the extent to which HPV-attributable cancer incidence varies across geographic regions, age groups, and birth cohorts in China is unclear. To address these gaps, we analysed the HPV-attributable cancer incidence in China using population-based cancer registry data from the CI5 database, covering six 5-year periods from 1988 to 2017. We evaluated temporal trends, provincial-level geographic variability, age-specific incidence patterns, and birth cohort effects and projected future incidence trends through 2032. By combining geographic variability analysis with age-period-cohort modelling, this study aimed to finely characterize the spatial and temporal heterogeneity of HPV-attributable cancer incidence in China to inform targeted cancer prevention and control strategies. Materials and methods Data sources Data on the incidence of HPV-associated cancers, including cervical (International Classification of Diseases Tenth Revision [ICD-10] code C53), vulvar (C51), vaginal (C52), penile (C60), anal (C21), laryngeal (C32), oral cavity (C02–06), and oropharyngeal (C01, C09–10) cancers, were retrieved from the CI5 Volumes VII–XII and stratified by sex and 5-year age group [9–14]. The CI5 database includes population-based cancer registries that undergo standardized quality assessment according to comparability, completeness, and validity [9–14]. In this study, we used only the registries included in the CI5 without additional selection. The selected study period (1988–2017) allowed for long-term assessments of HPV-attributable cancer ASR trends to capture the effects of major public health interventions, including improved cervical cancer screening. Population projections were sourced from World Population Prospects 2022 (United Nations) [15]. Data on the prevalence of high-risk HPV and the positivity rates of specific biomarkers associated with HPV-attributable cancers were retrieved from Chinese and English literature searches (details in the Supplementary Materials–Methods: meta-analysis). For biomarker selection, China-specific HPV E6/E7 mRNA positivity rates were applied to cervical squamous cell carcinoma (SCC), adenocarcinoma (ADC), and laryngeal cancers. For oral cavity cancer, pooled HPV E6/E7 mRNA positivity rates were estimated from studies conducted in China, East Asia and globally. For oropharyngeal cancer, combined \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\mathrm{p}\mathrm{1}{\mathrm{6}}^{\mathrm{INK4a}}$$\end{document} and HPV E6/E7 mRNA positivity rates specific to China were used in accordance with validated protocols [16]. Owing to limited data availability in China, the global \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\mathrm{p}\mathrm{1}{\mathrm{6}}^{\mathrm{INK4a}}$$\end{document} and/or HPV E6/E7 mRNA positivity rates were used for vulvar, vaginal, penile, and anal cancers [7, 17]. Statistical analysis HPV prevalence in relevant cancers and specific biomarker positivity rates among HPV-positive cases were integrated across the Chinese population via meta-analyses (results in Supporting Information). Briefly, differences among studies were evaluated through Cochran’s Q test and the I-square statistic (I2) [18]. According to heterogeneity, a fixed-effects model (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{I}^{2}<50\%$$\end{document}) or a random-effects model (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{I}^{2}\ge\:50\%$$\end{document}) was used to calculate the pooled HPV prevalence and specific biomarker positivity rates and their 95% confidence intervals (CIs) among HPV-positive cases [19]. The population-attributable fraction (PAF), an epidemiological metric that quantifies the aetiological contribution of causal factors to disease occurrence, was calculated according to the prevalence of infection with high-risk HPV (including HPV16, 18, 31, 33, 35, 39, 45, 51, 52, 56, 58, 59, 66, and 68) and the corresponding biomarker positivity rates among HPV-positive tumour cases [20]. Conceptually, the PAF represents the proportion of disease occurrence attributable to HPV infection under the specified attribution framework. Given the established causal role of persistent high-risk HPV infection in cervical cancer, we considered that 100% of overall cervical cancer cases were attributable to HPV infection [1]. For cervical SCC and ADC, however, the PAFs were empirically estimated with the available subtype-specific high-risk HPV prevalence and corresponding biomarker positivity data rather than being fixed at 100%. For HPV-attributable cancer at anatomical sites for which empirical HPV prevalence and biomarker data were available, the PAF was calculated as follows [7]:\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:PAF=P\times\:R$$\end{document} where P denotes the high-risk HPV prevalence among cancers at the corresponding anatomical site or histological subtype and R represents the corresponding biomarker positivity rate among high-risk HPV-positive cancer cases. For cancer sites with available genotype-specific HPV prevalence data, genotype-specific attributable fractions were estimated for each available high-risk HPV genotype using its corresponding HPV prevalence and the same genotype-nonspecific biomarker positivity rate. The incidence rates of HPV-attributable cancer were calculated by multiplying the incidence rates of HPV-related cancer by the corresponding PAF, stratified by sex, diagnostic period, and geographic location. Overall, HPV-attributable cancer incidence rates were calculated by multiplying the sex-specific incidence rates of HPV-associated cancers by the corresponding PAFs stratified by geographic location for 2013–2017 and by calendar period from 1988–1992 to 2013–2017. Owing to limited data availability, the cancer types included in the calculation of sex-specific combined counts of HPV-attributable cancer cases and incidence rates for 2013–2017 differed between the sexes. Among females, cervical, vulvar, vaginal, oropharyngeal, and laryngeal cancers were included in the data analyses. Among males, penile, oropharyngeal, oral cavity, and laryngeal cancers were included. In subsequent analyses, all reported results for HPV-attributable anal cancer and oral cavity cancer referred to estimates for both sexes combined, not sex-specific estimates. Uncertainty analysis The ASRs of HPV-attributable cancers per 100,000 were calculated using direct standardization with the world standard population, and the corresponding 95% CIs were calculated via a 100,000-run Monte Carlo simulation [21–23]. This procedure combines two uncertainty sources, the Poisson sampling error in cancer incidence rates and the meta-analytic uncertainty in PAF, into the final compound CIs. Geographical analysis For the geographical analyses, we included 157 Chinese cancer registries from the CI5 Volume XII (covering the diagnostic period 2013–2017) and grouped them into 25 provincial-level administrative regions according to their geographic location and the administrative division of the Ministry of Civil Affairs of China (Table S1, Supplementary Materials) [14, 24]. We performed Spearman correlation analysis to assess the associations between the Human Development Index (HDI) and the provincial-level ASRs of HPV-attributable cervical cancer, and a multivariable linear ecological regression analysis to adjust for hospital bed density and cancer registry population coverage. Temporal analysis Joinpoint regression model Joinpoint regression models were constructed to examine the temporal ASR changes from 1988 to 2017. The data were obtained from the aforementioned CI5 Volumes, which report cancer incidence in consecutive 5-year periods and provide reliable incidence rates with little random variation. Accordingly, the analysis was based on six time points rather than annual observations. Joinpoint regression was applied to these period-based ASRs to identify significant changes in temporal trends. Given the limited number of data points, models with 0 or 1 joinpoint were tested using the Monte Carlo permutation method to select the optimal segmentation [25]. The joinpoint annual percent change (JAPC) was then computed for each segment. Temporal trends were summarized with the average annual percent change (AAPC) and their 95% CIs across all segments [25]. For all ASRs, a significant increase was inferred when the 95% CI of the AAPC or JAPC was greater than 0; a significant decrease was inferred when the 95% CI of the AAPC or JAPC was less than 0; and a stable trend was inferred when the 95% CI of the AAPC or JAPC included 0. Synthetic birth cohorts Synthetic birth cohorts were derived by subtracting the midpoint age of each 5-year age group from the midpoint year of the corresponding 5-year calendar period. To evaluate the impact of HPV prevention and control measures on cancer incidence in specific age groups, age-specific incidence rates were plotted semilogarithmically against the birth cohort. The near-parallelism of age-specific curves across birth cohorts reflects the dominance of cohort effects on temporal trends. Age-period-cohort model Age-period-cohort (APC) models were constructed to evaluate the effects of age, period, and birth cohort on the HPV-attributable cancer incidence using age groups (a = 35–39, 40–44, ..., 80–84) and calendar periods (p = 1988–1992, ..., 2013–2017) [26]. A common age range of 35–84 years was applied across all cancer sites to ensure a consistent analytical framework and APC pattern comparability. For several noncervical HPV-attributable cancers, cases among individuals younger than 35 years were limited (including frequent zero cells), which could increase the random variation in age-specific incidence rates and compromise the stability and precision of APC estimates. Therefore, we used 35 years as a common lower age boundary rather than applying full available age ranges. Assuming that incident cases follow a Poisson distribution, we defined \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\mathrm{y}}_{ap}$$\end{document} as the number of incident cases and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\mathrm{n}}_{ap}$$\end{document} as the person-years at risk in age group a and period p according to the following equation:\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\mathrm{y}}_{ap}\sim\:\mathrm{Poisson}\left({\mathrm{n}}_{ap}\times\:{R}_{ap}\right)$$\end{document} The log-linear model was then used to model the log-transformed incidence rate as follows:\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mathrm{log}\left({R}_{ap}\right)=\mu+\alpha_{a}\mathrm{+}\beta_{p}\mathrm{+}\gamma_{\mathrm{c}}$$\end{document} where \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{R}_{ap}$$\end{document} denotes the incidence rate; \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\mu\:$$\end{document} is the intercept; and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\alpha_{a}, \beta_{p}, \gamma_c$$\end{document} represent the estimated age, period, and birth cohort effects, respectively. This framework quantifies the independent contributions of age effects (biological or behavioural changes across life stages), period effects (temporal influences affecting all ages, such as health policies), and birth cohort effects (risk variations from shared early-life exposures). Owing to the exact linear dependency of c=p-a, APC models exhibit inherent nonidentifiability, preventing simultaneous, unique estimations of all three effects. This issue was addressed by limiting the linear trend of the period effects to a zero mean and expressing the cohort effects as incidence rate ratios (IRRs) relative to a reference cohort (median number of cases) [26]. Bayesian age-period-cohort model Future HPV-attributable cancer incidence trends (from 2018–2022 to 2028–2032) were projected using Bayesian age-period-cohort (BAPC) models via the integrated nested Laplace approximation (INLA) framework [27], which yields faster Bayesian inference than Markov chain Monte Carlo methods [28]. In this study, the BAPC model was used to generate counterfactual projections by extrapolating historical age, period, and cohort patterns observed up to 2017, assuming their continuation beyond 2017. These projections did not incorporate potential future changes in risk factors, vaccination uptake, screening expansion, or diagnostic practices. The model specifications are as follows: \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\mathrm{log}\left({R}_{ap}\right)\ =\mu+\alpha_{a}\ +\ \beta_{p}\ +\ \gamma_c\ +\ \epsilon_{ap}$$\end{document}\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\epsilon_{ap}\sim\:\mathrm{N}\left(\mathrm{0,}\sigma_\epsilon^{\mathrm{2}}\right)$$\end{document} where \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\mu\:$$\end{document} is the global intercept; \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\alpha_{a},\beta_{p}\ \mathrm{and}\ \gamma_{\mathrm{c}}$$\end{document} (cohort c=p-a) denote the age, period, and cohort effects under second-order random walk (RW2)/ first-order random walk (RW1) priors, respectively; and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\mathrm{ε}}_{ap}$$\end{document} represents Gaussian overdispersion. The final model specifications are presented in Table S2. Gamma-distributed hyperpriors governed the precision parameters \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:({\uptau\:}=1/\sigma^{\mathrm{2}})$$\end{document} as follows: age, period, and cohort effects \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$(\tau_\alpha,\ \tau_\beta\ ,\tau_\gamma\sim\:\mathrm{Gamma}\left(\mathrm{1,\:0.0005}\right))$$\end{document} and overdispersion \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$(\tau_\epsilon\sim\:\mathrm{Gamma}\left(\mathrm{1,\:0.0005}\right))$$\end{document}. Sensitivity analysis To assess potential selection biases arising from both geographic coverage imbalances and temporal registry growth, we conducted two exploratory sensitivity analyses. First, to evaluate selection bias from urban overrepresentation and regional nonrepresentativeness in CI5 Volume XII (2013–2017), we performed post-stratification sensitivity analyses that adjusted for national urban-rural proportions and macro-regional population distributions [29]. Second, to evaluate selection bias resulting from the growth of included registries across successive CI5 volumes (Volumes VII to XII, 1988–2017), we performed a quantitative sensitivity analysis using the CI5 Plus dataset (1993–2017), which restricts data to a consistent set of 8 cancer registries with continuous longitudinal reports in China [30]. Details on the post-stratification weighting procedures, Joinpoint regression parameters, period-based segmentation, and statistical evaluation criteria for both sensitivity analyses are provided in the Supplementary Materials–Methods: Sensitivity analysis. A two-sided p value less than 0.05 was considered to indicate statistical significance. Because the analyses were primarily descriptive and exploratory, no adjustments for multiple comparisons were applied, and statistically significant findings from subgroup and exploratory analyses were interpreted cautiously. Joinpoint regression models were constructed using the Joinpoint Regression Program (version 5.4.0) [31]. All other data analyses and visualizations were conducted in R (version 4.5.0) [32]. Results HPV DNA prevalence, specific biomarker positivity rates, and PAFs varied substantially across cancer types in 2013–2017 (Table 1). Among cancer sites for which the HPV DNA prevalence was estimated from the literature, the highest was for cervical SCC (95.9%, 95% CI: 94.5%–97.3%) among females and for penile cancer (46.1%, 27.7%–64.4%) among males, whereas positivity for specific biomarkers was highest for cervical SCC and ADC (97.6%, 96.7%–98.5%) and oropharyngeal cancer (85.8%, 78.7%–90.8%). Accordingly, cervical cancer was assumed to be 100% attributable to HPV infection (as mentioned previously) and was therefore assigned a PAF of 100%, whereas the PAFs for other cancer sites ranged from 7.1% for laryngeal cancer among females to 38.6% for penile cancer among males. Table 1Input parameters and population-attributable fractions of HPV infection by cancer site, 2013–2017Cancer siteSexHigh-risk HPV prevalence(95% CI)Specific biomarker positivity rate(95% CI)PAF (95% CI)SourceCervix uteri†Females--100.0-Cervix uteri (SCC)Females95.9 (94.5, 97.3)97.6 (96.7, 98.5)93.6 (92.0, 95.2)HPV prevalence from Supporting Information Table S1 and Figure S2. Specific biomarker positivity rate from Table S5 and Figure S6.Cervix uteri (ADC)Females86.2 (76.7, 95.7)97.6 (96.7, 98.5)84.1 (74.8, 93.4)HPV prevalence from Supporting Information Table S2 and Figure S3. Specific biomarker positivity rate from Table S5 and Figure S6.AnusBoth sexes combined77.9 (65.8, 90.1)85.6 (82.0, 88.8)66.7 (56.0, 77.4)HPV prevalence from Supporting Information Table S6 and Figure S8. Specific biomarker positivity rate from de Sanjosé S et al., 2019[17].VulvaFemales46.9 (21.6, 72.3)85.9 (82.4, 88.8)40.2 (18.4, 62.1)HPV prevalence from Supporting Information Table S9 and Figure S12. Specific biomarker positivity rate from de Sanjosé S et al., 2019[17].VaginaFemales59.6 (51.9, 67.3)95.4 (92.4, 97.5)56.8 (49.3, 64.3)HPV prevalence from Supporting Information Table S12 and Figure S16. Specific biomarker positivity rate from de Sanjosé S et al., 2019[17].PenisMales46.1 (27.7, 64.4)83.8 (79.4, 87.6)38.6 (23.1, 54.1)HPV prevalence from Supporting Information Table S15 and Figure S20. Specific biomarker positivity rate from de Sanjosé S et al., 2019[17].OropharynxBoth sexes combined30.2 (21.7, 38.7)85.8 (78.7, 90.8)25.9 (18.4, 33.5)HPV prevalence from Supporting Information Table S18 and Figure S24 for both sexes combined, Table S19 and Figure S25 for males, and Table S20 and Figure S26 for females. Specific biomarker positivity rate from Supporting Information Table S22 and Figure S28.Males27.0 (13.9, 40.0)85.8 (78.7, 90.8)23.1 (11.8, 34.5)Females35.4 (7.3, 63.4)85.8 (78.7, 90.8)30.3 (6.1, 54.5)Oral cavity#Both sexes combined10.2 (5.1, 15.4)41.1 (18.1, 68.8)4.2 (1.3, 8.0)HPV prevalence from Supporting Information Table S23 and Figure S29 for both sexes combined, Table S24 and Figure S30 for males, and Table S25 and Figure S31 for females. Specific biomarker positivity rate from Supporting Information Table S27 and Figure S33.Males1.5 (0, 4.5)41.1 (18.1, 68.8)0.6 (0, 1.8)Females-41.1 (18.1, 68.8)-LarynxBoth sexes combined19.0 (13.0, 25.0)64.0 (32.6, 86.7)12.1 (6.2, 19.1)HPV prevalence from Supporting Information Table S28 and Figure S34 for both sexes combined, Table S29 and Figure S35 for males, and Table S30 and Figure S36 for females. Specific biomarker positivity rate from Supporting Information Table S32 and Figure S38.Males14.7 (4.3, 25.0)64.0 (32.6, 86.7)9.4 (2.5, 18.1)Females11.0 (0, 25.1)64.0 (32.6, 86.7)7.1 (0, 16.6)HPV human papillomavirus, CI confidence interval, PAF population-attributable fraction, SCC squamous cell carcinoma, ADC adenocarcinomaThe symbol “-” indicates that the value was not applicable and could not be calculated†For cervical cancer, high-risk HPV prevalence and specific biomarker positivity rate were not applicable because this study set the PAF to 100%#For other cancer sites, HPV prevalence data were unavailable; therefore, the PAF could not be estimated Table 1Input parameters and population-attributable fractions of HPV infection by cancer site, 2013–2017Cancer siteSexHigh-risk HPV prevalence(95% CI)Specific biomarker positivity rate(95% CI)PAF (95% CI)SourceCervix uteri†Females--100.0-Cervix uteri (SCC)Females95.9 (94.5, 97.3)97.6 (96.7, 98.5)93.6 (92.0, 95.2)HPV prevalence from Supporting Information Table S1 and Figure S2. Specific biomarker positivity rate from Table S5 and Figure S6.Cervix uteri (ADC)Females86.2 (76.7, 95.7)97.6 (96.7, 98.5)84.1 (74.8, 93.4)HPV prevalence from Supporting Information Table S2 and Figure S3. Specific biomarker positivity rate from Table S5 and Figure S6.AnusBoth sexes combined77.9 (65.8, 90.1)85.6 (82.0, 88.8)66.7 (56.0, 77.4)HPV prevalence from Supporting Information Table S6 and Figure S8. Specific biomarker positivity rate from de Sanjosé S et al., 2019[17].VulvaFemales46.9 (21.6, 72.3)85.9 (82.4, 88.8)40.2 (18.4, 62.1)HPV prevalence from Supporting Information Table S9 and Figure S12. Specific biomarker positivity rate from de Sanjosé S et al., 2019[17].VaginaFemales59.6 (51.9, 67.3)95.4 (92.4, 97.5)56.8 (49.3, 64.3)HPV prevalence from Supporting Information Table S12 and Figure S16. Specific biomarker positivity rate from de Sanjosé S et al., 2019[17].PenisMales46.1 (27.7, 64.4)83.8 (79.4, 87.6)38.6 (23.1, 54.1)HPV prevalence from Supporting Information Table S15 and Figure S20. Specific biomarker positivity rate from de Sanjosé S et al., 2019[17].OropharynxBoth sexes combined30.2 (21.7, 38.7)85.8 (78.7, 90.8)25.9 (18.4, 33.5)HPV prevalence from Supporting Information Table S18 and Figure S24 for both sexes combined, Table S19 and Figure S25 for males, and Table S20 and Figure S26 for females. Specific biomarker positivity rate from Supporting Information Table S22 and Figure S28.Males27.0 (13.9, 40.0)85.8 (78.7, 90.8)23.1 (11.8, 34.5)Females35.4 (7.3, 63.4)85.8 (78.7, 90.8)30.3 (6.1, 54.5)Oral cavity#Both sexes combined10.2 (5.1, 15.4)41.1 (18.1, 68.8)4.2 (1.3, 8.0)HPV prevalence from Supporting Information Table S23 and Figure S29 for both sexes combined, Table S24 and Figure S30 for males, and Table S25 and Figure S31 for females. Specific biomarker positivity rate from Supporting Information Table S27 and Figure S33.Males1.5 (0, 4.5)41.1 (18.1, 68.8)0.6 (0, 1.8)Females-41.1 (18.1, 68.8)-LarynxBoth sexes combined19.0 (13.0, 25.0)64.0 (32.6, 86.7)12.1 (6.2, 19.1)HPV prevalence from Supporting Information Table S28 and Figure S34 for both sexes combined, Table S29 and Figure S35 for males, and Table S30 and Figure S36 for females. Specific biomarker positivity rate from Supporting Information Table S32 and Figure S38.Males14.7 (4.3, 25.0)64.0 (32.6, 86.7)9.4 (2.5, 18.1)Females11.0 (0, 25.1)64.0 (32.6, 86.7)7.1 (0, 16.6)HPV human papillomavirus, CI confidence interval, PAF population-attributable fraction, SCC squamous cell carcinoma, ADC adenocarcinomaThe symbol “-” indicates that the value was not applicable and could not be calculated†For cervical cancer, high-risk HPV prevalence and specific biomarker positivity rate were not applicable because this study set the PAF to 100%#For other cancer sites, HPV prevalence data were unavailable; therefore, the PAF could not be estimated Input parameters and population-attributable fractions of HPV infection by cancer site, 2013–2017 HPV human papillomavirus, CI confidence interval, PAF population-attributable fraction, SCC squamous cell carcinoma, ADC adenocarcinoma The symbol “-” indicates that the value was not applicable and could not be calculated †For cervical cancer, high-risk HPV prevalence and specific biomarker positivity rate were not applicable because this study set the PAF to 100% #For other cancer sites, HPV prevalence data were unavailable; therefore, the PAF could not be estimated An estimated 75,631 HPV-attributable cancer cases occurred in China in 2013–2017 (Table 2). The burden of HPV-attributable cancer was substantially greater among females (68,955 cases; ASR = 11.20 per 100,000; 95% CI: 10.98–11.41) than among males (3463 cases; 0.52 [0.20–0.82]). Cervical cancer was the most common HPV-attributable cancer among females (ASR = 10.92 per 100,000; 95% CI: 10.83–11.00), whereas laryngeal and penile cancers were predominant among males (0.21 [0.03–0.37] and 0.18 [0.11–0.25], respectively). Notably, the ASRs for HPV-attributable oropharyngeal and laryngeal cancers were higher among males (0.12 per 100,000; 95% CI: 0.06–0.18 and 0.21 per 100,000 [0.03–0.37], respectively) than among females (0.05 per 100,000 [0.01–0.08] and 0.01 per 100,000 [0–0.03], respectively). Table 2Population-attributable fractions, numbers of cancer cases attributable to HPV, and age-standardized incidence rates of cancer attributable to HPV in China in 2013–2017Cancer siteBoth sexes combinedFemaleMalePAF (95% CI)Cases attributable to HPVASR† attributable to HPV (95% CI)PAF (95% CI)Cases attributable to HPVASR† attributable to HPV (95% CI)PAF (95% CI)Cases attributable to HPVASR† attributableto HPV (95% CI)Cervix uteri100.067,05310.92 (10.83, 11.00)---Anus66.7 (56.0, 77.4)21110.15 (0.12, 0.17)------Vulva40.2 (18.4, 62.1)8250.12 (0.05, 0.18)---Vagina56.8 (49.3, 64.3)6710.10 (0.09, 0.12)---Penis---38.6 (23.1, 54.1)12250.18 (0.11, 0.25)Oropharynx25.9 (18.4, 33.5)11440.09 (0.06, 0.11)30.3 (6.1, 54.5)3110.05 (0.01, 0.08)23.1 (11.8, 34.5)7830.12 (0.06, 0.18)Oral cavity4.2 (1.3, 8.0)6400.05 (0.01, 0.08)---0.6 (0, 1.8)560.01 (0, 0.02)Larynx12.1 (6.2, 19.1)19620.14 (0.07, 0.22)7.1 (0, 16.6)950.01 (0, 0.03)9.4 (2.5, 18.1)13990.21 (0.03, 0.37)Overall HPV-attributable burden-75,631*-#-68,95511.20 (10.98, 11.41)-34630.52 (0.20, 0.82)PAF population-attributable fraction, HPV human papillomavirus, ASR age-standardized incidence rate, CI confidence intervalThe symbol “-” indicates that the ASR attributable to HPV could not be estimated†Per 100 000 person-years*Overall HPV-attributable burden is the sum of the number of cases of site-specific HPV-attributable cancer, including anus, oropharynx, oral cavity, and larynx estimated using both-sex parameters; cervix uteri, vulva, and vagina estimated using female-specific parameters; and penis estimated using male-specific parameters#Overall HPV-attributable incidence rate was not calculated because the included cancer sites are associated with different underlying populations at risk and cannot be combined using a single denominator Table 2Population-attributable fractions, numbers of cancer cases attributable to HPV, and age-standardized incidence rates of cancer attributable to HPV in China in 2013–2017Cancer siteBoth sexes combinedFemaleMalePAF (95% CI)Cases attributable to HPVASR† attributable to HPV (95% CI)PAF (95% CI)Cases attributable to HPVASR† attributable to HPV (95% CI)PAF (95% CI)Cases attributable to HPVASR† attributableto HPV (95% CI)Cervix uteri100.067,05310.92 (10.83, 11.00)---Anus66.7 (56.0, 77.4)21110.15 (0.12, 0.17)------Vulva40.2 (18.4, 62.1)8250.12 (0.05, 0.18)---Vagina56.8 (49.3, 64.3)6710.10 (0.09, 0.12)---Penis---38.6 (23.1, 54.1)12250.18 (0.11, 0.25)Oropharynx25.9 (18.4, 33.5)11440.09 (0.06, 0.11)30.3 (6.1, 54.5)3110.05 (0.01, 0.08)23.1 (11.8, 34.5)7830.12 (0.06, 0.18)Oral cavity4.2 (1.3, 8.0)6400.05 (0.01, 0.08)---0.6 (0, 1.8)560.01 (0, 0.02)Larynx12.1 (6.2, 19.1)19620.14 (0.07, 0.22)7.1 (0, 16.6)950.01 (0, 0.03)9.4 (2.5, 18.1)13990.21 (0.03, 0.37)Overall HPV-attributable burden-75,631*-#-68,95511.20 (10.98, 11.41)-34630.52 (0.20, 0.82)PAF population-attributable fraction, HPV human papillomavirus, ASR age-standardized incidence rate, CI confidence intervalThe symbol “-” indicates that the ASR attributable to HPV could not be estimated†Per 100 000 person-years*Overall HPV-attributable burden is the sum of the number of cases of site-specific HPV-attributable cancer, including anus, oropharynx, oral cavity, and larynx estimated using both-sex parameters; cervix uteri, vulva, and vagina estimated using female-specific parameters; and penis estimated using male-specific parameters#Overall HPV-attributable incidence rate was not calculated because the included cancer sites are associated with different underlying populations at risk and cannot be combined using a single denominator Population-attributable fractions, numbers of cancer cases attributable to HPV, and age-standardized incidence rates of cancer attributable to HPV in China in 2013–2017 PAF population-attributable fraction, HPV human papillomavirus, ASR age-standardized incidence rate, CI confidence interval The symbol “-” indicates that the ASR attributable to HPV could not be estimated †Per 100 000 person-years *Overall HPV-attributable burden is the sum of the number of cases of site-specific HPV-attributable cancer, including anus, oropharynx, oral cavity, and larynx estimated using both-sex parameters; cervix uteri, vulva, and vagina estimated using female-specific parameters; and penis estimated using male-specific parameters #Overall HPV-attributable incidence rate was not calculated because the included cancer sites are associated with different underlying populations at risk and cannot be combined using a single denominator Marked geographical variability in the ASRs of HPV-attributable cancers was observed across the 25 provincial-level administrative regions covered by the 157 included registries during 2013–2017 (Fig. 1). Among females, the ASRs of HPV-attributable cervical cancer varied approximately 2.8-fold, from 6.15 per 100,000 (95% CI: 5.88–6.42) in Beijing to 17.43 per 100,000 (15.84–19.03) in Hunan (Supplementary Materials Table S13). Moreover, the ASRs of HPV-attributable cervical cancer and the provincial HDI values were negatively correlated (r= -0.50; P = 0.011) (Supplementary Materials Figure S1). This inverse association was further confirmed via multivariable ecological modelling after adjusting for hospital bed density and CI5 coverage (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\beta\:=$$\end{document}-3.02; 95% CI: -5.92–-0.12), wherein every 0.1 unit increase in provincial HDI was associated with a 26.07% (95% CI: 1.2%–44.7%) reduction in the ASR of HPV-attributable cervical cancer (Supplementary Materials Figure S2). Geographic estimates for individual cancer sites are provided in Supplementary Materials Tables S13–S22. Fig. 1Distribution of world age-standardized incidence rates (ASRs) of HPV-attributable cervical cancer, cervical squamous cell carcinoma (SCC), laryngeal cancer, and oropharyngeal cancer across provincial-level administrative regions in China during the 2013–2017 period Distribution of world age-standardized incidence rates (ASRs) of HPV-attributable cervical cancer, cervical squamous cell carcinoma (SCC), laryngeal cancer, and oropharyngeal cancer across provincial-level administrative regions in China during the 2013–2017 period From 1988 to 2017, the ASRs of HPV-attributable cervical, penile, and male oropharyngeal cancers significantly increased, with the steepest increase observed for oropharyngeal cancer among males (AAPC = 9.65%, 95% CI: 5.81–13.62) (Supplementary Materials Table S23, Fig. 2). In contrast, the ASRs of HPV-attributable laryngeal cancer decreased in both sexes over the entire period. The AAPC of HPV16-attributable cervical cancer was 8.31% (95% CI: 4.59–12.17), and the AAPC for HPV18-attributable cervical cancer was 7.97% (95% CI: 1.83–14.47) during 1988–2017 (Supplementary Materials Table S24 and Figure S3). Fig. 2Trends in world age-standardized incidence rates (ASRs) of HPV-attributable cancers in China from the periods 1988–1992 through 2013–2017. All the graphs are plotted on a semilogarithmic scale, and the rates are plotted at the midpoint of each 5-year interval. Solid dots indicate the observed ASRs. Lines represent fitted trends estimated from Joinpoint regression models. SCC, squamous cell carcinoma; ADC, adenocarcinoma Trends in world age-standardized incidence rates (ASRs) of HPV-attributable cancers in China from the periods 1988–1992 through 2013–2017. All the graphs are plotted on a semilogarithmic scale, and the rates are plotted at the midpoint of each 5-year interval. Solid dots indicate the observed ASRs. Lines represent fitted trends estimated from Joinpoint regression models. SCC, squamous cell carcinoma; ADC, adenocarcinoma The age-specific incidence patterns differed substantially across HPV-attributable cancers (Fig. 3). The incidence rates of HPV-attributable cervical ADC among females and HPV-attributable oropharyngeal cancer among males markedly increased across successive birth cohorts. Furthermore, there were increases in the incidence rates of HPV-attributable penile cancer and HPV-attributable vulvar and cervical cancers (including cervical SCC) in cohorts born before the late 1960s, followed by a slight decrease or stability in rates in younger cohorts. The incidence rates of HPV-attributable laryngeal cancer among females displayed considerable decreases across successive birth cohorts in the 35–79 age groups, whereas those among males sharply decreased in earlier cohorts, stabilized in middle cohorts, and slightly declined in younger cohorts. Fig. 3Incidence rates of HPV-attributable cancers by birth cohort and age group in China. All the graphs are plotted on a semilogarithmic scale. The incidence rates for 5-year age groups (i.e., 35–39 years to 80–84 years) are presented across birth cohorts. Birth cohorts were determined as the difference between calendar period and age at diagnosis, covering individuals born between 1908 and 1978 Incidence rates of HPV-attributable cancers by birth cohort and age group in China. All the graphs are plotted on a semilogarithmic scale. The incidence rates for 5-year age groups (i.e., 35–39 years to 80–84 years) are presented across birth cohorts. Birth cohorts were determined as the difference between calendar period and age at diagnosis, covering individuals born between 1908 and 1978 Adjusted for period effects, APC analysis revealed distinct cohort patterns across HPV-attributable cancers (Fig. 4). Among females, the pronounced increase in the IRR was observed for HPV-attributable cervical cancer, with progressively higher IRRs among cohorts born from the 1930s onwards, and a more gradual increase among cohorts born after the early 1960s. HPV-attributable vulvar cancer showed increasing IRRs among cohorts born after the 1940s. Among males, successive birth cohorts showed rapid increases in IRRs for HPV-attributable oropharyngeal cancer, whereas the IRRs of HPV-attributable penile cancer tended to increase among cohorts born between 1940 and 1960 but were stable for younger cohorts. In contrast, the cohort-specific IRRs for HPV-attributable laryngeal cancer tended to decrease, whereas those for oral cavity and anal cancers were relatively stable or decreased. Notably, the estimates for cohorts born before the 1930s were associated with greater uncertainty because of limited observations, as reflected by substantially wider 95% CIs for these extreme IRRs (particularly for cervical SCC, vulvar, and oropharyngeal cancers among females and oropharyngeal and penile cancers among males). Fig. 4Age and birth-cohort effects estimated from age-period-cohort models for HPV-attributable cancers in China. The left panels show estimated age-specific incidence rates, and the right panels show incidence rate ratios (IRRs) associated with birth cohorts from 1908 to 1978. Shaded areas represent 95% confidence intervals of the estimated effects. All the graphs are plotted on a semilogarithmic scale. Birth cohorts were determined as the difference between the calendar period and ages at diagnosis Age and birth-cohort effects estimated from age-period-cohort models for HPV-attributable cancers in China. The left panels show estimated age-specific incidence rates, and the right panels show incidence rate ratios (IRRs) associated with birth cohorts from 1908 to 1978. Shaded areas represent 95% confidence intervals of the estimated effects. All the graphs are plotted on a semilogarithmic scale. Birth cohorts were determined as the difference between the calendar period and ages at diagnosis Under the counterfactual continuation of historical incidence trends through 2017, the ASR of HPV-attributable cervical cancer was projected to increase by 78.8% (8.61/10.92), from 10.92 per 100,000 (95% CI: 10.83–11.00) in 2013–2017 to 19.53 per 100,000 (95% uncertainty interval [UI]: 9.99–34.36) in 2028–2032 (Supplementary Materials Table S32, Fig. 5). In contrast, an increase (73.3%) was projected for the ASR of HPV-attributable anal cancer in both sexes combined, from 0.15 per 100,000 (95% CI: 0.12–0.17) in 2013–2017 to 0.26 per 100,000 (95% UI: 0.06–0.74) in 2028–2032. The greatest increase (258.3%) was projected for HPV-attributable oropharyngeal cancer among males, from 0.12 per 100,000 (95% CI: 0.06–0.18) in 2013–2017 to 0.43 per 100,000 (95% UI: 0.18–0.80) by 2028–2032. Fig. 5Projected world age-standardized incidence rates (ASRs) of cancers attributable to HPV infection during 2018–2032. All the graphs are plotted on a semilogarithmic scale, with rates shown at the midpoint of each 5-year interval. Filled circles indicate the observed ASRs, and the solid line indicates the projected rates. The shaded fan represents the uncertainty interval, with the outer boundaries corresponding to the 2.5th and 97.5th percentiles. The vertical dashed line indicates the beginning of the projection period (2018). Projections represent counterfactual scenarios assuming continuation of historical age, period, and cohort patterns observed through 2017 Projected world age-standardized incidence rates (ASRs) of cancers attributable to HPV infection during 2018–2032. All the graphs are plotted on a semilogarithmic scale, with rates shown at the midpoint of each 5-year interval. Filled circles indicate the observed ASRs, and the solid line indicates the projected rates. The shaded fan represents the uncertainty interval, with the outer boundaries corresponding to the 2.5th and 97.5th percentiles. The vertical dashed line indicates the beginning of the projection period (2018). Projections represent counterfactual scenarios assuming continuation of historical age, period, and cohort patterns observed through 2017 Sensitivity analyses were performed to evaluate the impact of geographic coverage imbalances and longitudinal registry growth across HPV-attributable cancer sites (Supplementary Materials Tables S35 and S36; Figures S4–S6). In the macroregion post-stratification sensitivity analysis, the relative differences between weighted and unweighted ASRs were small for HPV-attributable cervical, vulvar, and oral cavity cancers among females, and HPV-attributable penile, oral cavity, oropharyngeal and laryngeal cancers among males (Supplementary Materials Table S35). Substantial relative differences were observed for HPV-attributable laryngeal, oropharyngeal, and vaginal cancers among females and for anal cancer in both sexes combined. In the longitudinal trend sensitivity analysis that compared the full dataset comprising 157 Chinese registry dataset in the CI5 volumes with a dataset of 8 registries spanning up to 25 consecutive years, long-term trend estimates revealed no statistically significant differences between the two datasets for HPV-attributable cervical cancer, including ADC and SCC, and laryngeal cancers among females and for oropharyngeal cancer among males, with the 95% CIs of the AAPC differences including zero (Supplementary Materials Table S36 and Figures S4–S6). Significant differences in long-term trend estimates were observed only for HPV-attributable penile and laryngeal cancers among males and for HPV-attributable oropharyngeal cancer in both sexes. Discussion This study provides a comprehensive overview of the incidence of HPV-attributable cancers in China from 1988 to 2017. In 2013–2017, cervical cancer accounted for the greatest burden among females, whereas head and neck cancers predominated among males, with substantial geographic variability observed across Chinese cancer registry populations. Over the past three decades, the incidence of several HPV-attributable cancers, particularly cervical cancer among females and oropharyngeal cancer among males, has increased substantially. Assuming the continuation of historical incidence patterns observed through 2017, the ASRs of these cancers are projected to continue to increase through 2028–2032. APC analysis further revealed pronounced birth cohort patterns, particularly for HPV-attributable oropharyngeal cancer among males and cervical cancer among females. During the 2013–2017 period, cervical cancer contributed to the greatest burden of HPV-attributable cancers in China, whereas oropharyngeal cancer predominated in developed countries such as the United States, which may be related to differences in cervical screening and HPV vaccination coverage [33]. The incidence of HPV-attributable head and neck cancers was higher among males in China, which is consistent with global patterns and with previous evidence suggesting that the increasing burden of oropharyngeal cancer may be related to an increase in oral HPV infection and other changes among younger people [34]. Together, these findings highlight the importance of considering HPV-attributable cancers beyond cervical cancer when the future cancer burden associated with HPV infection is evaluated. The substantial geographic heterogeneity observed in HPV-attributable cancer incidence should be interpreted carefully. The provincial HDI was inversely associated with HPV-attributable cervical cancer incidence; however, this ecological association should not be interpreted as a direct causal relationship. Differences in cervical cancer screening coverage, HPV vaccination uptake, healthcare accessibility, HPV exposure, diagnostic capacity, and cancer registration completeness may have contributed to the observed regional variability [35]. In particular, the higher recorded incidence in economically developed regions may partly reflect greater screening coverage, improved diagnostic capacity, and more complete cancer registration rather than a higher baseline incidence alone. Conversely, the lower recorded incidence in resource-limited areas may partly reflect underdetection or incomplete cancer registration [36]. These contextual factors may also help explain the lack of uniformity in the relationship between the HDI and the cervical cancer incidence across provinces. Although HPV vaccination uptake may contribute to future regional differences in the HPV-attributable cancer burden, its impact is unlikely to be fully reflected in our study because HPV vaccination was introduced in mainland China only after 2016 [5]. Therefore, the observed ecological association should be interpreted as reflecting a combination of socioeconomic, healthcare-related, behavioural, and surveillance-related factors rather than socioeconomic development itself as a direct determinant of the HPV-attributable cervical cancer incidence. The geographic variability should also be interpreted in the context of the composition of the CI5 registries, which undergo standardized quality control procedures but do not reflect a probability sample of the Chinese population [37]. Registry coverage differs across provinces, and urban and eastern populations are more strongly represented than rural and western populations [38]. Differences in registry composition, population coverage, diagnostic practices, screening efforts, and registration completeness may therefore contribute to the observed geographic heterogeneity. The results of our post-stratification sensitivity analysis suggest that the estimates for major cancers, particularly those for cervical cancer and head and neck cancer, were relatively robust after adjusting for the observed geographic and urban-rural composition of the registry coverage. In contrast, greater differences were observed for less common malignancies, specifically anogenital cancers. These findings increase our confidence in the primary estimates for cervical and head and neck cancers but indicate greater uncertainty and the need for more cautious interpretation for anogenital cancers. Importantly, post-stratification reweighting quantifies the magnitude of regional bias without removing underlying registration biases. The incidence of HPV-attributable cervical cancer increased over the last three decades; similar increasing patterns were observed for HPV-attributable anogenital and oropharyngeal cancers. These findings are broadly consistent with previous reports of the increasing incidence of HPV-attributable cancers in China [35]. However, the magnitude of the projected increase differs between the studies, which may be explained by differences in study period, data sources, cancer sites included, and methods used to estimate HPV-attributable fractions. Notably, the projected increase in the incidence of HPV-attributable anal cancer reflects potential future challenges associated with HPV-related cancers beyond cervical cancer. Although universal screening for anal cancer has not been established, continued surveillance remains important for high-risk groups [39]. The historical incidence data used in this study largely predate the widespread implementation of HPV vaccination in China. The first HPV vaccine was approved for use in mainland China in 2016, and HPV vaccination was not incorporated into the national immunization programme until November 2025. Given the limited time between vaccine introduction and the end of our observation period (2017), the population-level effects of vaccination on cancer incidence would not be expected to be detectable in these historical data. Accordingly, the BAPC estimates in the present study represent counterfactual projections based on the continuation of historical incidence patterns rather than forecasts incorporating future changes in HPV vaccination, screening, treatment, or other interventions. The birth cohort pattern for HPV-attributable cervical cancer, including both SCC and ADC, showed a marked increase in risk among cohorts born before the early 1960s, followed by a relative attenuation in younger cohorts. This pattern may reflect multiple sources, including generational differences in HPV exposure and the phased implementation and expansion of cervical cancer screening in China [4]. Changes in screening practices may have altered the detection and incidence patterns across successive generations, making it difficult to attribute the cohort pattern to changes in HPV exposure alone. Despite the relative attenuation observed among younger generations, the persistence of the absolute burden indicates that HPV-attributable cervical cancer is likely to remain an important public health challenge in the foreseeable future [40]. For HPV-attributable vulvar cancer, the cohort-specific risk increased among those born from the 1940s onwards, potentially reflecting generational differences in HPV exposure [41]. HPV-attributable penile cancer showed a similar increase in risk among cohorts born between 1940 and 1960, followed by a more gradual increase in younger cohorts. Such patterns may reflect generational differences in HPV exposure, sexual behaviours, hygiene practices, circumcision occurrence, and other unmeasured factors [42]. In the absence of population-based screening methods for vulvar cancer, appropriate vulvar clinical examinations along with cervical cancer screening remain important for early disease detection [43]. A pronounced sex disparity was observed in the cohort effects for HPV-attributable oropharyngeal cancer. Among females, older generations exhibited a higher baseline risk of HPV-attributable oropharyngeal cancer, while the risk across successive generations among males increased more rapidly and surpassed that for females born after the early 1950s. These patterns may reflect sex-specific differences in HPV exposure, but other factors should also be considered. Smoking and alcohol consumption are established risk factors for head and neck cancers and may vary across generations, potentially contributing to the observed cohort patterns [44]. Temporal changes in diagnostic capabilities, HPV detection, and histopathological classification may also have influenced the identification and attribution of HPV-related oropharyngeal cancers [45]. In addition, temporal changes in HIV prevalence may have influenced the persistence of HPV infection and cancer risk through immunosuppression among populations at higher risk of HIV infection [46]. However, the extent to which any of these factors contributed to the observed cohort effects cannot be determined with the present ecological data. Overall, the observed cohort patterns likely represent the joint influence of changes in HPV exposure; coexisting risk factors, including smoking and alcohol consumption; temporal changes in HIV prevalence among populations at high risk of HIV infection; and healthcare-related factors, such as diagnostic capabilities and histopathological classification. These findings highlight the complexity of interpreting APC patterns in HPV-attributable cancers. The findings of this study have several important implications for preventing and controlling HPV-attributable cancers in China. Given the projected increase in the incidence of several HPV-attributable cancers, improving both primary prevention through HPV vaccination and secondary prevention through organized screening programmes is essential [47]. Expanding these strategies could not only reduce the future burden of HPV-attributable cancers but also alleviate the long-term healthcare and economic burden associated with cancer diagnosis, treatment, and survival. Because HPV-related cancers are largely preventable, vaccination and screening investments are likely to generate substantial long-term health gains while reducing avoidable healthcare expenditures. The results of previous modelling studies have further suggested that HPV vaccination combined with cervical cancer screening would be cost-effective and could substantially accelerate progress towards eliminating cervical cancer [48]. The pronounced geographic heterogeneity reported in this study further suggests that prevention and healthcare resources should be prioritized according to regional disease burden, screening capacity, healthcare accessibility, and opportunities for improving prevention coverage, particularly in areas with persistently high or rapidly increasing incidence rates [49, 50]. Such an approach may improve resource allocation efficiency while addressing inequalities in access to HPV-attributable cancer prevention and care. Additionally, the increasing burden of HPV-attributable cancers, especially oropharyngeal cancer among males, suggests that the public health benefits of HPV prevention extend beyond cervical cancer. Although vaccination of females currently remains the priority, our findings support further evaluation of strategies for preventing HPV-associated cancers among males, including sex-neutral vaccination strategies, within the context of disease burden, healthcare resources, and local public health priorities [51]. Collectively, these measures would contribute to achieving the World Health Organization’s 90–70–90 targets and the “Healthy China 2030” goals [52]. Taken together, the results of this study add to the existing evidence on HPV-attributable cancers in China by combining registry-based geographic variation with long-term temporal and birth cohort analyses. Rather than providing a national estimate of the HPV-attributable cancer burden, our study characterizes heterogeneity across Chinese cancer registry populations, identifies age- and cohort-specific patterns underlying temporal changes, and provides counterfactual projections according to the continuation of historical incidence patterns. This integrated approach provides a highly detailed epidemiological characterization of HPV-attributable cancers in China and may help identify populations and regions in which prevention and healthcare resources require greater attention. Nevertheless, this study had several data-related and methodological limitations. First, the HPV-attributable cancer estimates were based on PAFs derived from HPV prevalence and biomarker positivity rates. Owing to the limited availability of high-quality and population-based biomarker data in China, we used globally aggregated data for some cancers, which may not fully reflect the Chinese epidemiological context. Second, because the CI5 registries may not fully represent the overall cancer burden in entire provinces, the ecological associations should be interpreted with caution. Although post-stratification sensitivity analyses were conducted to evaluate the robustness of the estimates to differences in registry composition, these analyses cannot eliminate the effects of unmeasured selection mechanisms or residual ecological confounding. Third, the APC analysis was limited to individuals aged 35–84 years to ensure a consistent age range across cancer sites and minimise instability arising from sparse case numbers in younger age groups. This restriction precluded the assessment of age-specific incidence patterns among individuals younger than 35 years and older than 84 years and narrowed the observable range of birth cohorts. In particular, for cervical cancer and other cancers in which incidence changes may occur at younger ages or among younger generations, excluding young individuals may attenuate or obscure emerging age-cohort patterns. Therefore, the APC findings should be interpreted as descriptions of patterns among individuals aged 35–84 years rather than across the entire age spectrum. Fourth, the BAPC projections were based on the continuation of historical age, period, and cohort patterns observed through 2017 and did not explicitly incorporate the potential effects of HPV vaccination programmes, expansion of cervical cancer screening, or other future interventions. Therefore, these estimates should be interpreted as counterfactual projections rather than forecasts of actual future incidence. Finally, incomplete pathological classification and insufficient HPV genotyping data precluded further stratified analyses. Future studies incorporating more representative Chinese biomarker data, broader registry coverage, individual-level determinants, and intervention-informed projection models are needed to refine the estimates of the HPV-attributable cancer burden and improve the interpretation of temporal and geographic variability. In summary, the epidemiological profile of HPV-attributable cancers in China is characterized by substantial heterogeneity across cancer types, sexes, regions, and generations. HPV-attributable cervical cancer continues to account for the greatest burden among females, while the burden of the HPV-attributable head and neck cancer among males is increasing, reflecting distinct but convergent public health challenges. The observed geographic and generational heterogeneity demonstrates the need for prevention strategies that can respond to differences in disease burden, healthcare access, and population risk. These findings provide epidemiological evidence to inform the prioritization of HPV vaccination, cervical cancer screening, cancer surveillance, and healthcare resources in a given context while demonstrating the need for further evaluations of broader HPV prevention strategies for males. Supplementary Information Supplementary Material 1. Supplementary Material 1. Supplementary Material 2: Supporting Information Supplementary Material 2: Supporting Information