Cancer burden attributable to risk factors, 1990–2019: A comparative risk assessment

Summary An up-to-date comprehensive assessment of the cancer burden attributable to risk factors is essential for cancer prevention. We analyzed the population attributable fraction (PAF) of cancer disability-adjusted life years (DALYs) attributable to 11 level 2 risk factors using data from the Global Burden and Disease Study (GBD) 2019. We highlighted that almost half of the cancer DALYs can be preventable by modifying relevant risk factors. The attributable cancer DALYs increased by 60.42%–105.0 million from 1990 to 2019. Tobacco, dietary risks, alcohol use, high body-mass index, and air pollution were the top five risk factors. The PAFs attributable to high fasting plasma glucose, high body-mass index, and low physical activity have increased worldwide from 1990 to 2019. Unsafe sex was the leading risk factor for women before age of 54. Tailored prevention programs targeted at specific populations should be scaled up to reduce the cancer burden in the future.


INTRODUCTION
Cancer remains a major threat to global human health worldwide.Among 22 groups of diseases and injuries, cancer was reported as the second leading cause of death and disability-adjusted life years (DALYs) after cardiovascular disease. 1 It was estimated that cancer was responsible for 23.6 million new cases, 10.0 million deaths, and 250 million DALYs globally in 2019, with increases in cancer incidence, mortality and DALYs of 26.3%, 20.9%, and 16.0%, respectively, from 2010 to 2019. 2 In addition, the World Health Organization predicted that global cancer incidence will increase to 24 million by 2035. 3 Primary prevention, early diagnosis and early treatment are the three major weapons against cancer.However, due to the limited medical resources and the high cost of therapies, people in many countries, including subpopulations in very high-income countries, have no access to comprehensive and systematic curative treatment for cancer.To date, therapeutic interventions have not significantly reduced overall cancer mortality, although some progress has been made for specific cancers. 4In this case, primary prevention has been identified as the optimal choice for reducing the global cancer burden.As reported by the World Cancer Research Fund, about one-third to one-half of all cancers may be preventable by modifying known environmental, lifestyle, and infection-related risk factors. 5evertheless, low budgets have been allocated to cancer prevention efforts even in Europe and the United States, due to the long cycle of cancer development, which requires time to see the benefits. 6revious epidemiological studies have reported the cancer burden attributable to potentially modifiable risk factors.][9][10][11] In addition, most studies defined cancer burden mainly as morbidity or mortality, ignoring the additional health burden such as living with disability, making it difficult for policymakers to prioritize resource allocation.3][14][15] A recent GBD study led by the GBD 2019 Cancer Risk Factors Collaborators reported the cancer mortality and DALYs attributable to potentially modifiable risk factors.Although there were some references to more detailed risk factors, most of this study focused on the cancer burden attributable to level 1 risk factors.

National attributable burden by SDI
The PAF also varied considerably among countries (Table S2; Figure 3).In 2019, the PAF of cancer DALYs (more than 40%) attributable to combined modifiable risk factors was higher in more developed locations, mainly in North America, Asia, and Europe.In contrast, the PAF was lower (less than 40%) in less developed areas, mainly in Africa, South America, and Oceania.Alarmingly, the PAF exceeded 50% in Serbia, Hungary, Kiribati, Montenegro, and Greenland.), respectively.Notably, tobacco was the leading risk factor in 166 countries, whereas unsafe sex was the leading risk factor in the remaining 38 countries, mainly concentrated in Africa (Table S2).
Figure S3 shows the correlations between the PAF attributable to selected risk factors and SDI in 204 countries and territories in 2019.The PAFs for other environmental risks, tobacco, alcohol use, drug use, high FPG, high BMI, low physical activity, and occupational risks were positively correlated with SDI, whereas the PAFs for air pollution and unsafe sex were negatively correlated with SDI.No significant correlation was observed between dietary risks and SDI.Attributable burden by age group and sex

ll OPEN ACCESS
In general, the age-specific PAF followed a unimodal distribution for most level 2 risk factors, with the highest PAF observed in middle-aged and older individuals.However, the PAFs for dietary risks and low physical activity increased or remained stable with increasing age in both men and women.(Figure 4).The PAF for air pollution peaked in the 65-69 age group for both men and women, with men having higher PAFs than women.For high BMI, the PAF was highest in men aged 50-54 years and in women aged 60-64 years.The PAF was significantly higher in men in the early age, while this pattern was reversed from the 50-54 age group onwards, as a steep increase was observed in women in this age group.For dietary risks, men had a higher PAF than women in the early age groups, whereas women's PAF exceeded that of men from the age group 75-79.For low physical activity, women had a higher PAF than men in all age groups.For occupational risks, men aged 85-89 years had the highest PAF, while women aged 70-74 years had the highest PAF, with men being consistently higher than women.The PAF attributable to unsafe sex reached a plateau in women aged 35-39 years.Notably, unsafe sex surpassed tobacco as the leading risk factor for cancer DALYs in women before the age of 54.

Attributable burden of specific cancer types across SDI quintiles
The PAF of type-specific cancer DALYs attributable to risk factors across SDI quintiles is shown in Table S3 and Figure 5. Specifically, tracheal, bronchus, and lung cancer was the only cancer type attributable to air pollution and other environmental risks.Sixteen tumor types were related to tobacco, with the highest PAF found in tracheal, bronchus, and lung cancer at the global level.However, it was worth noteworthy that the PAF for larynx cancer exceeded that for tracheal, bronchus, and lung cancer in high and high-middle SDI quintiles.Alcohol use was associated with 8 categories of tumors, with the highest PAF observed for lip and oral cavity cancer, nasopharynx cancer, and other pharynx cancers.Liver cancer was the only one cancer type associated with drug use.High FPG was correlated with 7 cancer types, with the highest PAF observed in bladder cancer.There were 13 categories of neoplasms associated with high BMI, with the highest PAF observed in uterine cancer.Dietary risks were correlated with five cancer types, with the highest PAF observed in colon and rectum cancer.Low physical activity was associated with breast cancer and colon and rectum cancer.Occupational risks were associated with seven cancers, with the highest PAF observed for mesothelioma.Cervical cancer was associated with unsafe sex, with a PAF of 100% observed for females worldwide.The PAFs of type-specific cancer DALYs attributable to risk factors were generally higher in higher SDI regions.However, there were exceptions.For example, the PAF attributable to dietary risks differed substantially between SDI quintiles, with the highest PAF observed for esophageal cancer and tracheal, bronchus, and lung cancer in lower SDI quintiles, and for breast cancer in higher SDI quintiles (Table S3; Figure

DISCUSSION
In 2019, there were 105.0 million cancer DALYs or 41.23% of total cancer DALYs attributable to all level 2 risk factors, with tobacco, dietary risks, alcohol use, high BMI, and air pollution being the top five risk factors.In contrast to the relatively stable trend in PAF for combined risk factors, the PAFs of cancer DALYs attributable to high FPG, high BMI, and low physical activity have been increasing in recent years, due to the economic development and transformation of lifestyle worldwide.
A recent GBD study led by the GBD 2019 Cancer Risk Factors Collaborators reported the cancer deaths and DALYs attributable to potential risk factors. 16Because of the same set of data used, most of our findings were comparable to theirs.However, their study only showed the geographical patterns and age distribution of the cancer burden attributable to level 1 risk factors, despite delineation on the sex disparity and SDI patterns of the cancer burden attributable to level 2 risk factors.To raise awareness of these level 2 risk factor estimates, our study highlighted the disparity in sex, age, locations, SDI, and cancer types in terms of cancer burden attributable to level 2 risk factors.
Development status was a significant contributor to cancer burden.The overall PAF was higher in higher SDI regions, suggesting that a higher proportion of cancer burden can be reduced in higher SDI regions if the exposure levels of relevant modifiable risk factors are controlled to the optimal exposure levels.For each specific risk factor, SDI was positively correlated with the PAFs of cancer DALYs attributable to other environmental risks, tobacco, alcohol use, drug use, low physical activity, occupational risks, high FPG and high BMI; negatively correlated with air pollution and unsafe sex; and unrelated to dietary risks.However, development status can only partially explain the variation in PAF across countries, given the complex and non-linear relationship between PAF and SDI.Considerable variations were observed in PAFs among countries, underscoring the need for finer delineation of high-risk populations.
Significant age and sex disparities were observed in the cancer burden attributable to risk factors.Men had a higher PAF attributable to combined risk factors than women, due to a significant contribution from the male-prominent risk factor-tobacco.To be specific, women PAFs and rates of cancer DALYs attributable to eleven level 2 risk factors at the regional level (A) Age-standardized PAFs of cancer DALYs attributable to eleven level 2 risk factors, for both sexes in 27 global and regional locations in 2019.(B) Age-standardized DALY rates of cancer attributable to eleven level 2 risk factors, for both sexes in 27 global and regional locations in 2019.DALYs, disabilityadjusted life years.PAF, population attributable fraction.See also Table S1.had higher PAFs attributable to wealth-related risk factors, including high BMI, high FPG, and low physical activity, and unsafe sex than men.Men had higher PAFs attributable to most behavioral, environmental, and occupational risk factors.The highest PAFs were observed in the middle-aged and older adults.Recognizing the long cycle of cancer development, we promoted early prevention to reduce cancer burden.In addition, the sex prominence for a specific risk factor was not consistent in all age groups.For example, men had higher PAF attributable to high BMI before age of 50.However, this pattern was reversed from the 50-54 age group onwards, with a steep increase in the PAF in this age group for women.These findings suggest the need for age-and sex-specific assessments prior to future cancer risk factor interventions.
Tobacco included smoking, secondhand smoke, and chewing tobacco.Tobacco was the leading risk factor in much of the world.Despite a notable decline in tobacco prevalence observed in much of the world, it was still on the rise in many less developed countries, such as Afghanistan, Albania, Saudi Arabia, Lebanon, and Mongolia. 17Central Europe and East Asia had the highest cancer death and DALY rates of cancer attributable to tobacco smoking. 13As of July 2019, the Tobacco-Free Initiative's MPOWER policy package has successfully expanded evidence-based tobacco control measures to nearly two-thirds of the world's population, including raising tobacco taxes, banning tobacco advertisement, and building a smoke-free environment. 18In response to this call, every country should spare no effort to accelerate the smoke-free process in the coming decades.
Although alcohol use and tobacco are the two most recognized carcinogens, efforts to reduce alcohol use have been relatively less proactive than for tobacco. 19As a result, the PAF of cancer DALYs attributable to alcohol use has increased over the observed periods, especially for men in high-income countries.A related GBD study also found that the cancer burden attributable to alcohol use has decreased from 1990 to 2019, and the attributable burden was higher in males, elders, and developed regions (based on SDI). 15Besides, another study reported that an estimated 741 300 or 4.1% of all new cancer cases were attributable to alcohol consumption in 2020, using cancer incidence data from GLOBOCAN 2020. 20Because we chose DALYs as the measurement, rather than incidence, the findings of this study were not comparable to ours.Considering a small amount of drinking is likely to be beneficial for cardiovascular function, the World Cancer Research Fund recommends no more than 2 drinks for men and 1 drink for women. 21wo forms of air pollution were quantified as contributing to the cancer burden in GBD 2019: ambient particulate matter pollution and household air pollution (HAP) from the use of solid fuels.Despite a large reduction in the prevalence of HAP, the prevalence of ambient particulate matter pollution has increased significantly since the 1990s 22 The GBD study reported that cancer deaths attributable to ambient particulate matter pollution increased by more than 300% from 1990 to 2017. 23Although China has recently experienced reductions in emissions from fossil fuel and solid biofuel, China still had the highest burden from ambient particulate matter pollution. 24Given the undesirable situation, the energy, industry, construction, and transport sectors should take action to curb the worsening global air pollution.For example, European legislation on controlling pollution from power generation, manufacturing, and road transport could prevent 163,000 deaths worldwide each year. 25ifteen dietary risk factors were identified by GBD 2019.A relevant study reported that almost 30%-35% of tumors were related to diet. 26It was estimated that that for every 50g increase in processed meat intake, the risk of developing cancer increased by 17%.Besides, for every 100g increase in red meat intake, the risk increased by 18%. 27On the contrary, regular intake of fresh fruits and vegetables can reduce the risk of stomach cancer, colorectal cancer, and breast cancer. 28As more and more countries have adopted the Western dietary style with a large amount of red meat, progressed meat, and insufficient intake of fruits and vegetables, it was in urgent need to construct a healthier diet with easy access to healthy fruits, vegetables, legumes, and whole grains and strict restrictions on the consumption of red meat, processed meat, and sugary drinks worldwide.Several promising interventions include: mass media campaigns, food and menu labeling, food pricing strategies (subsidies and taxes), and dietary legislations et al. 29 The attributable burden of metabolic risk factors has been increasing worldwide in recent years, particularly high FPG and high BMI.Notably, women were more susceptible to the metabolic risk factors along with low physical activity, because a few metabolic-related cancers such as uterine cancer and breast cancer were prominent in women.A more pronounced gender gap was observed after the age of 54 years, suggesting that these metabolic risk factors pose a greater threat to postmenopausal women than premenopausal women. 30The American Cancer Society (ACS) estimates that at least 20% of incident cancers are attributable to obesity, and obese patients had a worse prognosis compared with patients with a normal BMI. 31 The increasing prevalence of high FPG and high BMI and the determinants behind require further investigation.However, it was universally acknowledged that lack of exercise, excessive intake of calories, undesirable diet quality, and diet composition were closely related to these metabolic risk factors. 32,33A multifactorial means was urgently needed to intervene these metabolic risk factors, such as reducing the intake of sugar, fat, and total calories, and promoting physical activity and exercise to the general public. 34nsafe sex was the leading risk factor in 38 countries and territories, most of which were in Africa.However, the impact of unsafe sex on the cancer burden has not been adequately assessed in previous studies.In our study, unsafe sex was the second leading risk factor responsible for cancer DALYs among women, but surpassed tobacco as the leading risk factor among women before the age of 54.Based on evidence in the literature, GBD 2019 attributed 100% of cervical cancers to unsafe sex.These sources stated that human papillomavirus (HPV) infection, particularly HPV-16 and HPV-18, was necessary for the development of cervical cancer and that HPV was spread only through sexual contact. 35gure 3. PAFs of cancer DALYs attributable to eleven level 2 risk factors at the national level Age-standardized PAF of cancer DALYs attributable to all modifiable risk factors (A) and eleven specific level 2 risk factors: air pollution (B), other environmental risks (C), tobacco (D), alcohol use (E), drug use (F), high fasting plasma glucose (G), high body-mass index (H), dietary risks (I), low physical activity (J), occupational risks (K), unsafe sex (L), for both sexes in 204 countries and territories in 2019.DALYs, disability-adjusted life years.PAF, population attributable fraction.See also Table S2 and Figure S3.
In recent years, vaccines against HPV-16 and HPV-18 have been approved for use in adolescents and young women worldwide.However, due to the high cost of developing sufficient vaccines, implementation of vaccines against HPV has not been sufficiently enforced in many less developed countries in sub-Saharan Africa and South Asia. 36,37There is an urgent need to strengthen sexual safety education and expand HPV vaccination among these countries.S3.
Infection was an important un-estimated risk factor in this study, due to the lack of data on the cancer burden attributable to infections in GBD 2019.The PAF of incident cancers attributable to infections was 16% globally in 2008, and higher in low-income countries (22.9%) than in high-income countries (7.4%). 38Owing to the poor sanitation and shortage of vaccines, countries in sub-Saharan Africa had the highest PAF (32.7%).Therefore, it is imperative to develop regional frameworks to increase vaccine coverage and improve public health in these target areas in the future. 5tudies have shown that prevention strategies targeting modifiable risk factors are cost-effective for cancer prevention.In addition, because of the interaction between risk factors, interference on one risk factor may influence the other one, which could further enhance the effect of risk factor intervention on cancer control. 39,40Epidemiological studies have shown that the effect of individual behavioral intervention is minimal, whereas societal actions such as regulatory controls on risk factors will have a substantial impact on cancer control. 41ppropriate approaches include: promoting the significance of primary prevention, adopting and implementing relevant legislations, and further investigating and enforcing the framework for reducing major carcinogens. 42Late diagnosis and scarce medical resources remains the major concerns in many less developed countries.Therefore, it is suggested that these countries should strengthen preventive measures along with early screening and palliative care services to reduce cancer burden. 43n conclusion, our study provided insight into the magnitude of the cancer burden attributable to 11 level 2 risk factors in 2019 and analyzed the trends over the past decades.We highlighted that a large proportion of cancer DALYs are preventable, with tobacco, dietary risks, alcohol use, high BMI, and air pollution being the top five risk factors.Tobacco remained the leading risk factor in 166 countries and unsafe sex in 38 countries, mainly in Africa.Unsafe sex was considered the female-specific risk factor in GBD 2019, accounting for most cancer DALYs in women before the age of 54.Due to the overwhelming increase in the prevalence, wealth-related risk factors, including high FPG, high BMI, and low physical activity, warranted policy attention worldwide.Our findings may provide insightful information for policymakers to establish cost-effective prevention programs targeting modifiable risk factors to reduce cancer burden in the future.

Limitations of the study
While there have been significant improvements in data collection and modeling in GBD 2019, the inherent defects of the GBD study should not be ignored.First and foremost, high-quality primary data are scarce in many less-developed countries, where the final estimates must rely on statistical modeling.Even when data are available, the discrepancies in data quality and accuracy can lead to the deviations in the estimates.Secondly, the RRs for each risk factor evaluated may vary across ethnic groups and geographical locations, in which case the use of a universal effect size would reduce the authenticity and reliability of the estimates.Thirdly, prevalence of risk factors and cancer DALYs were obtained in the same year, regardless of the lag time between risk factor exposure and cancer development. 44Thus, for a selected risk factor, its estimated PAF may have been overestimated when it was rising and underestimated when it was declining during the observed periods.Lastly, GBD 2019 provided relevant data only when there was solid evidence of a causal relationship between selected risk factors and cancer.Future GBD rounds should explore the attributable cancer burden for other potential risk factors.

Risk-outcome pairs
A total of 481 risk-outcome pairs were selected as meeting the evidence grading criteria of being convincing or probable provided by the World Cancer Research Fund. 46These risk factors were divided into four hierarchies, with level 1 representing the overarching categories including behavioral, environmental, occupational and metabolic; level 2 included 20 individual risks or risk clusters; level 3 risk factors were the individual risk disaggregated from risk clusters within level 2; level 4 detailed risks in the most disaggregated manner. 22The GBD risk hierarchy with levels is presented in Table S4.We extracted 11 level 2 risk factors related to cancer, including air pollution, other environmental risks, tobacco, alcohol use, drug use, high FPG, high BMI, dietary risks, low physical activity, occupational risks, and unsafe sex.Twenty-three cancers were associated with these 11 risk factors.Table S5 shows the International Classification of Diseases (ICD) codes mapped to the cancer mortality data in this study.

Exposure estimation
The exposure data come from many sources, including authoritative systematic reviews, large cohort studies, household surveys, national censuses, and satellite data.To make the exposure data from different sources more consistent and suitable for modelling, adjustments

Figure 1 .
Figure 1.PAFs of cancer DALYs attributable to eleven level 2 risk factors at the global level The temporal trends of global age-standardized PAFs of cancer DALYs attributable to eleven level 2 risk factors for both sexes (A), male (B) and female (C) from1990 to 2019.DALYs, disability-adjusted life years.PAF, population attributable fraction.See also Figures S1 and S2.

Figure 2 .
Figure 2. PAFs and rates of cancer DALYs attributable to eleven level 2 risk factors at the regional level (A) Age-standardized PAFs of cancer DALYs attributable to eleven level 2 risk factors, for both sexes in 27 global and regional locations in 2019.(B) Age-standardized DALY rates of cancer attributable to eleven level 2 risk factors, for both sexes in 27 global and regional locations in 2019.DALYs, disabilityadjusted life years.PAF, population attributable fraction.See also TableS1.

Figure 4 .
Figure 4. Age and sex disparity in the PAFs of cancer DALYs attributable to eleven level 2 risk factors at the global level Global age-specific PAF of cancer DALYs attributable to eleven specific level 2 risk factors for male(A) and female(B) in 2019.DALYs, disability-adjusted life years.PAF, population attributable fraction.See also TableS1.

Figure 5 .
Figure 5. PAFs of cancer DALYs attributable to eleven level 2 risk factors and related cancer types across SDI quintiles in 2019 DALYs, disability-adjusted life years.PAF, population attributable fraction.SDI, sociodemographic index.See also TableS3.

Table 1 .
Global estimates of number, rate, and PAF of cancer DALYs attributable to eleven level 2 risk factors and the percent change by sex from 1990 to 2019

Table 1 .
Continued The highest PAFs for both other environmental risk factors and tobacco were observed in the 65-69 age group for men and in the 70-74 age group for women, with men higher than women.The PAF for alcohol use peaked in the 45-49 age group for men and in the 50-59 age group for women.For drug use, the PAF was highest in the 55-59 age group for men and in the 65-69 age group for women, with men being higher up to the 60-64 age group and women reversing this trend from the 65-69 age group onwards.For high FPG, the highest PAF was observed in the 75-79 age group for men and in the 80-84 age group for women, with women having consistently higher PAF except in the 50-54 age group.
or RESOURCE SOURCE IDENTIFIER