The comprehensive methodology of the GBD 2021 study has been extensively detailed in previous publications (10–12). Burden data was sourced from the Institute for Health Metrics and Evaluation (IHME). This involved obtaining essential metrics such as counts, rates, and age-standardized rates for incidence (ASIR), prevalence (ASPR), mortality (ASMR), and DALYs (ASDR), from 1990 to 2021. Female cancers were standardized utilizing the International Classification of Diseases (ICD) codes (10). To evaluate the female cancer burden in China, two primary data sources were utilized: surveillance data from the China Disease Surveillance Points (DSP) system and virtual registry (VR) datasets compiled by the Chinese Center for Disease Control and Prevention (CDC) (10). The GBD 2021 study is publicly available with anonymous data; thus, ethics committee review is not required. The flowchart of data source, screening, and statistical analysis was presented in Supplementary Figure S1.

The burden of female cancers is typically quantified through the following methods: (1) The original data from two primary data sources underwent a data processing process that mainly includes standardization, mapping data to GBD causes, age/sex splitting, and redistribution (10). (2) Subsequently, two concurrent selection processes were initiated: one aimed at generating input data for the mortality-to-incidence rate (MIR) model, while the other was focused on generating incidence to obtain a final mortality estimate (10). (3) Combined incidence, mortality data, and MIRs to generate mortality estimates. Concurrently, cancer incidence was calculated from cancer mortality estimates using MIRs (10). (4) The cause of death (CoD) database was created, and the years of life lost (YLLs) were assessed using the cause of death ensemble model (CODEm) (10). (5) Based on the GBD incidence estimates and the survival dataset, the final prevalence estimates were evaluated and were split into four sequelae (11). (6) For disability estimation, sequelae-specific years lived with disability (YLDs) were estimated using their specific weights. The disease model meta-regression (DisMod-MR 2.1) was used, and total YLDs were the sum of these sequelae-specific YLDs (11). (7) DALYs were estimated by summing YLLs and YLDs (11).

The GBD risk factor analysis used the comparative risk assessment (CRA) framework to compute estimates, typically involving the following process: (1) For each risk-outcome pair, relative risks (RRs) were individually estimated (12). (2) Estimating exposure levels and distributions of each risk factor using Bayesian models: spatiotemporal Gaussian process regression (ST-GPR) and DisMod-MR 2.1 (12). (3) Based on epidemiological evidence, determine the theoretical minimum risk exposure levels (TMRELs) (12). (4) determined the population attributable fraction (PAF) (12). (5) For each risk, the summary exposure values (SEVs) were computed (12). (6) Mediation factors were assessed to adjust PAF overestimation due to risk factor independence and were used to calculate the burden from multiple risk factor interplay (12). (7) Multiplying PAFs by DALYs for a specific outcome yields measures of attributable burden (12, 13).

Data was described with absolute numbers, rates, and 95% uncertainty intervals (UIs). The UIs were derived by utilizing the 2.5th and 97.5th percentiles from a distribution comprising 1,000 draws for each respective metric (11). To assess temporal trends, we utilized the estimated annual percentage change (EAPC) as a metric (14, 15). This was derived using a regression formula: ln(age-standardized rate) = α + β × year + ε. Using this formula, we computed the EAPC and its corresponding 95% confidence interval (CI) via the expression [100 × (exp(β) − 1)] (14, 15).

We utilized a framework based on the APC model to explore potential trends in the burden of female cancer, taking into account factors such as age, period, and birth cohort (16), via the online tool at http://analysistools.nci.nih.gov/apc/. In forecasting the disease burden of female cancers, we employed the Bayesian APC (BAPC) model for predictive analysis, projecting the disease burden up to 2040. (17). This model is formulated as N_ij = log(λ_ij) = μ + α_i + β_j + γ_k, where λ_ij represents the case count, μ is the intercept, and α_i, β_j, and γ_k, respectively, indicate the influences of age, period, and cohort (17). To address overdispersion, we implemented the BAPC model through the INLA and BAPC packages in R software (17).

Statistical analyses were done using R software (version 4.4.0), and results were visualized, with significance determined at a two-tailed p-value threshold of < 0.05.

In 2021, the estimated number of newly diagnosed female breast cancer patients in China was 38.58 × 10⁴, corresponding to ASIR with around 37.00 per 100,000 population. The number of prevalence cases and ASPR of female breast cancer were 374.81 × 10⁴ and 355.72 per 100,000. Breast cancer contributed to 8.81 × 10⁴ deaths among the female population, and ASMR was 8.24 per 100,000. An estimated 292.11 × 10⁴ DALYs in 2021 were attributed to female breast cancer, with the ASDR at 281.54 per 100,000 population (Table 1).

The number of incidence cases and ASIR of cervical cancer in 2021 were 13.28 × 10⁴ and 13.37 per 100,000 among the female population. The prevalence of cervical cancer is estimated to be 74.65 × 10⁴ with an ASPR of 79.03 per 100,000. About 4.98 × 10⁴ people died of cervical cancer in 2021, accompanied by ASMR with around 4.64 per 100,000. Cervical cancer caused 154.78 × 10⁴ DALYs, and ASDR was 149.84 per 100,000 (Table 1).

The estimated number of new cases of uterine cancer was 7.20 × 10⁴ with an ASIR of 6.65 per 100,000. The prevalence of uterine cancer was reported to be 50.07 × 10⁴ cases with an ASPR of 46.52 per 100,000. Uterine cancer contributed to 1.36 × 10⁴ deaths, and ASMR was 1.24 per 100,000. Additionally, uterine cancer resulted in 40.55 × 10⁴ DALYs with an ASDR of 37.86 per 100,000 (Table 1).

There were an estimated 4.12 × 10⁴ new cases of ovarian cancer, with an ASIR of 4.05 per 100,000. The prevalence of ovarian cancer was estimated to be 17.82 × 10⁴ cases with an ASPR of 18.67 per 100,000. Ovarian cancer contributed to 2.51 × 10⁴ deaths, and ASMR was 2.30 per 100,000. Besides, an estimated 75.05 × 10⁴ DALYs were attributed to ovarian cancer, with an ASDR of 71.16 per 100,000 (Table 1).

The incidence rates for breast and uterine cancer peaked in the 60–64 age group, cervical cancer in the 55–59 age group, and ovarian cancer in the 65–74 age group. The prevalence rates for breast cancer peaked at 60–64 years, cervical and ovarian cancer at 50–54 years, and uterine cancer at 55–59 years. The highest mortality rates were observed in those over 95 years old for breast and cervical cancer, and in the 90–94 age group for uterine and ovarian cancer. The peak DALYs rates were seen at 55–59 years for breast and cervical cancer, 65–69 years for uterine cancer, and 60–64 years for ovarian cancer (Figure 1; Supplementary Table S1). Overall, the diagnoses of female cancers predominantly occurred between the ages of 55–74 years, with the highest burden of DALYs rates observed in the ages of 55–69 years.

From 1990 to 2021, the numbers of incidence, prevalence, mortality, and DALYs of breast cancer significantly increased among the female population. The ASIR and ASPR of breast cancer doubled from 1990 to 2021 (EAPC = 2.34, and 2.49, respectively) (Supplementary Figure S2; Table 1). The increasing trend was observed across most age groups, with the most pronounced increase in the 60–64 age group (Supplementary Figures S4, S5; Supplementary Table S1). The ASMR and ASDR for breast cancer present an overall declining trend (EAPC = −0.62, and −0.52, respectively), but exhibited an upward trend from 2014 to 2021 (Supplementary Figure S2). The decline in DALYs rate and mortality rate from 1990 to 2021 was observed across all age groups, except for the 60–64 age group (Figure 2 and Supplementary Figure S3; Supplementary Table S1).

For cervical, uterine, and ovarian cancer, the numbers of incidence, prevalence, mortality, and DALYs increased slowly from 1990 to 2021. The ASIR of cervical cancer (EAPC = 0.88) and uterine cancer (EAPC = 0.40) all presented a slight increasing trend, while the ASIR of ovarian cancer showed a slight declining trend (EAPC = −0.44) (Table 1; Supplementary Figure S2). Notably, the incidence rates for ovarian cancer declined in most age groups, except for an increase noted in the 20–34 age group (Supplementary Figure S4; Supplementary Table S1). The mortality and DALYs rates of cervical, uterine, and ovarian cancer showed a consistent downward trend across all age groups (Figure 2 and Supplementary Figures S2, S3).

Age effects of the incidence and prevalence rates for breast cancer exhibit a sharp increase with age, particularly after the age of 40, peaking in the 85–89 age group, followed by a decline and fluctuation. Age effects in the mortality rate of breast cancer showed a linear increase, and the rates surged after age 85. The DALYs rate for breast cancer increases with age, peaking in the 55–60 age group, then declining until age 90, followed by a slow rise. As for cervical, uterine, and ovarian cancer, the incidence and prevalence rates gradually increase with age, peaking in the 55–60 age group, and then decreasing slowly. The mortality rate of cervical cancer shows a linear increase with age, while the mortality rates for uterine and ovarian cancer show a gradual increase with age, followed by a slight decrease after age 80. The DALYs rates of cervical, uterine, and ovarian cancer all peak in the middle-aged group of 55–60, with a subsequent decline (Figure 3A).

Period effects generally showed a significantly increasing risk of incidence and prevalence for breast cancer over the study period, while a generally declining risk of mortality and DALYs, with a notable resurgence post-2012. Cervical cancer experienced its lowest period risk of incidence between 1997 and 2001, followed by a gradual increase. Conversely, mortality and DALYs rates for cervical cancer showed a significant decline after the period of 2002–2006. The incidence and prevalence rates of uterine cancer reached their peak in the period of 2006–2011, while the mortality and DALYs rates exhibited a clear downward trend over the study period. Period effects on the incidence, mortality, and DALYs for ovarian cancer displayed a downward trend after the period of 1997–2001, followed by a resurgence post the period of 2012–2016 (Figure 3B).

Cohort analyses reveal unfavorable trends in the incidence and prevalence of breast cancer among more recent cohorts, while a gradually declining risk of mortality and DALYs. The incidence and prevalence rates of cervical cancer are elevated in older cohorts but have decreased in more recent ones, indicating a possible success in prevention or early detection. Unlikely, the incidence and prevalence rates for uterine and ovarian cancer have remained relatively stable across all birth cohorts. Cohort effects on mortality and DALYs rates for cervical cancer and ovarian cancer followed similar declining trends observed in breast cancer, while uterine cancer exhibited a more pronounced decrease in recent cohorts (Figure 3C).

High red meat consumption was identified as the predominant contributor to DALYs of breast cancer, accounting for approximately 13% of the total burden, with a stable contribution across all years and age-specific cohorts (Figure 4). The attributable percentage of DALYs for breast cancer due to high BMI and elevated FPG has shown a gradual increase over the years. Other risk factors, such as alcohol use, low physical activity, secondhand smoke, and smoking, constitute a smaller proportion of the DALYs, remaining generally stable over time but exhibiting variations across different age groups. Notably, low physical activity has a more significant impact on older age populations. Nearly all DALYs for cervical cancer, across all age groups and over the years, were attributable to unsafe sex practices. Additionally, 5–7% of these DALYs were attributed to smoking, particularly in individuals aged 55 years and older. It is noteworthy that the proportion of DALYs for uterine and ovarian cancer attributable to high BMI has increased over the years. For uterine cancer, this proportion rose from 11.56% in 1990 to 26.90% in 2021, and for ovarian cancer, it increased from 1.17% in 1990 to 6.81% in 2021. Besides, the proportion of DALYs for ovarian cancer attributable to occupational asbestos exposure remained stable over the years and was highest in the older population, specifically those over 70 years old (Figure 4). Risk-attributable female cancer mortality analysis revealed consistent results (Supplementary Figure S6).

Based on predictions, by 2040, the numbers of female breast cancer incident cases, prevalent cases, deaths, and DALYs would increase to 87.08 × 10⁴, 830.28 × 10⁴, 16.22 × 10⁴, and 472.22 × 10⁴, respectively; the corresponding ASIR, ASPR, ASMR, and ASDR would also increase to 64.38, 604.11, 9.86, and 347.15 per 100,000, respectively (Figure 5; Supplementary Figure S7; Supplementary Table S2). For cervical cancer, the ASIR, ASPR, ASMR, and ASDR would decrease to 11.07, 61.50, 2.98, and 98.50 per 100,000, respectively. For uterine cancer, the ASIR, ASPR, ASMR, and ASDR would also increase to 10.79, 76.55, 1.47, and 43.23 per 100,000, respectively. Similar trends were also identified in ovarian cancer burden, with the ASIR, ASPR, ASMR, and ASDR increasing to 5.46, 25.32, 3.01, and 90.66 per 100,000, respectively (Figure 5; Supplementary Table S2).