Global Magnitude and Temporal Trend of Mesothelioma Burden Along with the Contribution of Occupational Asbestos Exposure in 204 Countries and Territories From 1990 to 2019: Results from the Global Burden of Disease Study 2019

Background Understanding the annual incidence, mortality, and disability-adjusted life-years (DALYs) for mesothelioma globally provides essential foundations for cancer control, policy decisions and resource allocation. methods Complying with the Global Burden of Disease Study 2019, we extracted the data of incidences, deaths and DALYs in 204 countries and territories from 1990 to 2019. Average annual percentage changes were used to quantify the temporal trends in mesothelioma burden. Besides, the population attributable fractions of the risk factor of mesothelioma were also estimated.


Introduction
Malignant mesothelioma is an aggressive, treatment-resistant cancer and its median survival time ranges from 4 to 12 months from diagnosis [1][2][3] . It is largely caused by asbestos exposure with an estimated proportion at 80% or more, other possible causes or contributing factors include family history, related germline gene mutations (such BAP1 and BML), ionizing radiation and exposure to other mineral bers such as erionite bers in Turkey [4][5][6][7] . It has been estimated worldwide that 30870 new cases and 26278 new deaths of mesothelioma occurred in 2020, with huge regional disparity 8 . Mesothelioma is commonly diagnosed at older ages as its latency period is typically longer than 30 years, and for which the incidence and mortality rates are much higher in male individuals than female individuals [9][10] . There is no curative treatment for mesothelioma, with systemic treatment options including chemotherapy, radiotherapy and targeted therapy. Current unimodal or multimodal treatment is usually of little bene t 2 11 . A large proportion of cases have a well-established causal relationship between exposure to asbestos and the development of malignant mesothelioma, while recent studies revealed that the cessation of exposure may not reduce risk in exposed individuals 12 . Therefore, reasonable and effective prevention programs can better reduce the mesothelioma burden and medical expenses. By 1990, the use of asbestos in most industrialized countries had been reduced by at least 75% from the peak asbestos consumption. Iran, Korea, Chile, and Egypt reached the same level of reduction of asbestos usage in 1999, as did Nigeria, Zimbabwe, the United Arab Emirates, Ukraine, and Kazakhstan between 2000 and 2005 13 . The huge geographic disparity and poorly characterized epidemiologic trends represent the complexity in the intervention of mesothelioma 14 . This highlights the need for the latest spatial distribution and temporal trends of mesothelioma across the world, which are the basic prerequisites for policy decisions and resource allocation. Researchers have presented the estimation of mesothelioma burden of 195 countries and territories based on the study of Global Burden of Disease, Injuries, and Risk Factors (GBD) 2017, which found that incident cases and deaths of mesothelioma continuously increased worldwide, especially in resource-limited regions with low sociodemographic index (SDI) levels 15 .
The GBD study 2019 is the most recent comprehensive and comparable data set evaluating epidemiologic levels and trends of 369 diseases along with 87 risk factors globally, which has incorporated new datasets, enhanced method performance and standardization, therefore superseding previously published GBD estimates [16][17] . For example, compared with GBD 2017, the mortality-toincidence ratio estimates in GBD 2019 were updated using lower case inclusion criteria and different model hyperparameters, resulting in more training data and less smoothing across time and geography. Nevertheless, there has been no study analyzing the latest mesothelioma burden and trends based on the GBD study 2019 until now. Therefore, the aim of this study was to use data from the GBD 2019 study to determine the global, regional, and national burdens of mesothelioma across 204 countries and territories between 1990 and 2019 by age, sex, and SDI. The ndings would help increase the world's attention to mesothelioma, as well as design speci c strategies for the prevention and intervention of mesothelioma for different regions respectively.

Data source
The exhaustive study framework and analysis methodology of the GBD 2019 study have been depicted in previous articles by the GBD collaborators 16 18 . The estimation of cancer burden in the GBD study across the world was calculated based on many original national cancer registry systems and the combined Asia, and Australasia (Table 1). Moreover, the world was also simpli ed into ve regions based on the corresponding SDI, namely, low, low-middle, middle, high-middle, and high SDI regions. The SDI could re ect the national healthcare development based on a country's lag-distributed income per capita, average years of schooling, and the total fertility rate in females under the age of 25 years 16 . According to the well-established inclusion criterion in GBD 2019 Study for a risk-outcome pair, only occupational exposure to asbestos was judged to have su cient evidence to prove a causal association with mesothelioma from potential behavioral, environmental and occupational, and metabolic risk factors.
The disease burden attributable to potential risk factors was assessed via the GBD comparative risk assessment framework, which includes the assessment of risk-outcome pairs, relative risks, theoretical minimum risk exposure level, and population attributable fraction 17 .

Statistical Analyses
To eliminate the difference of the age composition of the populations, the age-standardized incidence rate (ASIR), age-standardized mortality rate (ASMR) and age-standardized DALY rate (ASDR) were calculated to compare the difference among different geographical areas, historical periods, and sexes, according to the standard age composition of GBD world population 20 . The average annual percentage change (AAPC) was widely used to estimate the overall temporal trend in burden rate of disease, which was derived via the following regression model, ln (burden rate)= α + β* calendar year + ϵ, and the AAPC with its 95% con dence interval (CI) were calculated based on the formula of 100 × (exp (β) − 1) 19 21-22 . In our study, the lower boundary of the 95%CI more than 0 indicated an increasing trend, conversely, the upper boundary of its 95%CI less than 0 indicated a decreasing trend. Otherwise, the burden rate was deemed to be relatively stable over the study period.
In addition, to explore the changing pattern of AAPCs in the burden rates of mesothelioma in 204 countries and territories, we applied the Spearman rank correlation method to calculate the overall correlation coe cients between the AAPCs and ASRs in 1990 as well as SDI in 2019, because of the nonnormal distribution of data. The ASRs of mesothelioma in 1990 could re ect the baseline burden of mesothelioma and the SDI in 2019 could denote the healthcare development status of each region. The relationship between the AAPC and ASRs of mesothelioma in 1990 was to test whether the high-burden countries paid more attention to the prevention and control of mesothelioma burden, and further identify high-burden countries with an overall upward trend. Taking into account the possible non-linear relationship, the Locally Weighted Scatterplot Smoothing (LOWESS) regression was applied to present more detailed information between the AAPC and possible factors 21 23 , which was performed using the geom_smooth function with default parameters of the ggplot2 package. The research analysis was implemented using the R program (version 4.0.3). A two-sided P value of less than 0.05 was considered statistically signi cant.

Results
Global burden and temporal trend in mesothelioma  Figure 1).
The incident cases of mesothelioma rose from age 50 years and remained relatively high level until age 90 years, with a peak at age 70-74 years among both sexes in 2019 (Figure 3, Figure S2). The incidence rate per 100000 rose to a peak at age 85-89 years and dropped substantially afterward in 2019 (Figure 3, Figure S2). The mortality rate reached a peak and remained high over age 85 years ( Figure S2-S3). The DALY rate increased signi cantly over age 60 years and held up highest among age 75-90 years ( Figure  S2, Figure S4). The difference of the incidence rate, mortality rate and DALY rate in ve SDI regions focused on the population aged over 60 years, and high SDI regions account for the highest proportion ( Figure 3, Figure S2-S4).
Globally, the incidence rate, mortality rate and DALY rate all decreased among the population aged under 70 years from 1990 to 2019, especially among higher SDI regions, though increased among the population aged over 80 years (Figure 3, Figure S3-S4).

The correlation between SDI and mesothelioma burden
We  Figure S5B, Figure S6B). Further, we analyzed the relationship between ASRs and SDI over time in 21 GBD regions, expressed in the annual time series of 1990 and 2019. During these 30 years, the ASRs in most GBD regions remained relatively stable, except Southern Sub-Saharan Africa, Andean Latin America, Eastern Europe, Western Europe, High-income Asia Paci c presented a climbing trend beforehand and following decline afterward; Australasia rebounded after the rst declining and has been through another declining in recent years; Southern Latin America, Central Europe presented a climbing trend and remained stable recently; while High-income North America presented a decreasing trend over time ( Figure 4C, Figure S5C, Figure S6C).

Risk Factors
The ASDR of mesothelioma attributable to occupational exposure to asbestos in 2019 was positively associated with SDI in 2019 at the national levels (ρ= 0.3105, P = 6.2e-06) ( Figure 5A). Most countries and territories have remained relatively constant or changed moderately between 1990 and 2019 for the proportion of DALYs due to occupational exposure to asbestos, except the Georgia increased markedly from 1990 to 2019 ( Figure 5B). The high-income GBD regions always showed the largest proportion of DALYs due to occupational exposure to asbestos, with most territories of high-income GBD regions reaching up to more than 90% ( Figure 5B).

Discussion
This study provides the latest and comprehensive assessment of the global burden of mesothelioma. From 1990 to 2019, the number of mesothelioma incident cases and deaths globally increased by nearly two times, while the age-standardized rates of incident cases, deaths and DALYs showed a slightly declining trend. The incidence rate of mesothelioma increased signi cantly in the 1960s due to the massive use of asbestos during World War II and thereafter, especially in those high-resource countries with advanced industries 24 . Governments came to recognize that asbestos is one of the most important occupational carcinogens and announced the necessity of decreasing industrial use of asbestos, so as to reduce the burden of asbestos-related disease 13 . Since the late 1970s and early 1980s, asbestos bans and regulations were implemented in most areas, bringing about a declining trend of the incidence rates afterwards 25 . There can be several reasons for the continuous increase in the absolute number of incident cases and deaths during the three decades. Firstly, due to the long latency period from asbestos exposure to the development of mesothelioma, the mesothelioma populations are usually diagnosed at elderly 26-27 ; Secondly, other carcinogenic mineral bers has been increasingly used freely and resulted in unexpected environmental exposure 6 28 ; Moreover, germline mutations and ionizing radiation indicating other susceptible subgroups for mesothelioma 25 .
Consistent with other worldwide studies, the incidence and mortality of mesothelioma were more than three times in males than in females globally across all age groups from 1990 to 2019. The ASIR, ASMR and ASDR decreased among females in the 30 years while remained relatively stable among males. The lower incidence and declining trend of mesothelioma in females might be partly owing to rare industrial labor and consequent less exposure to asbestos 24 . As previously reported, there is a 35-40 years latency period from asbestos exposure to the development of mesothelioma, consequently, the incident cases and deaths would increase with the aging of the population. In this study, we observed that the number of incident cases and deaths gradually increased with age and reached a peak in the elderly aged 70-90 years. Furthermore, the global incidence rate, mortality rate and DALY rate all decreased among the population aged under 70 years from 1990 to 2019, however increased among the population aged over 80 years. The elderly patients with mesothelioma are generally refractory because of the extent of disease, advanced age, comorbidities or poor performance status, therefore are usually considered for palliative treatment instead. Thus, early detection of mesothelioma, multi-channel interventions and novel targeted treatments are urgently needed to be developed. Of note, recently available data support a possible role of miRNAs and DNA methylation in the etiology of mesothelioma and suggest their potential as biomarkers for the early diagnosis and even for the screening of an at a high-risk population of asbestos-exposed individuals [29][30][31][32] . Besides, several novel therapeutic agents are under investigation in recent years, such as targeted therapies and immunotherapies, which may provide further treatment options and improve clinical outcomes for mesothelioma in the future [33][34] .
During the past three decades, ASIR has been higher in high and high-middle SDI regions and lower in low SDI regions. The high incidence rate of mesothelioma in the higher SDI regions may be related to the previous industrial use of asbestos. Although the limitation and bans of asbestos have been implemented earlier in industrialized countries, it would take a long time to observe the decrease of incident rates due to the more than 30 years of the latency period. It is supposed that mesothelioma cases might begin to decrease after 2 decades of a complete ban on asbestos use. In this 30-year crosssectional study, we observed a declining trend of ASIR in the high-middle and high SDI regions, as well as ASMR and ASDR. This illustrates that the asbestos regulation earlier in those industrialized regions is beginning to take effect. For the middle, low-middle and low SDI regions, the incident cases, deaths and DALYs were much lower compared with high SDI regions. The possible reason is that the resource-limited regions with small industrial volumes are related to comparatively few asbestos occupational exposure.
Besides, the diagnosis rate and data availability of mesothelioma have been a matter of concern in developing regions 35 . Recent articles reported a very high rate of incorrect diagnoses, ranging from approximately 14% in the western world to approximately 50% in some developing countries, which can in uence all the incidence, mortality and DALYs statistics 36 . The ASRs in the middle and low-middle SDI regions showed a rising trend, especially the low-middle SDI regions showed the highest increasing trend of ASRs, which might be partly owing to the later asbestos regulation, and also the gradually improved diagnosis rate and data accessibility.
The AAPC of age-standardized burden rates were all negatively correlated with baseline ASIR, ASMR and ASDR in 1990 at the national level, partially because the statistics change were more apparent in the countries with low burden and small base in 1990. Moreover, data analysis showed no statistical correlation between SDI and AAPC values in ASIR, ASMR, and ASDR in 204 countries or territories. This is different from the estimation of mesothelioma burden of 195 countries and territories based on GBD 2017, which showed the AAPC of incidence rate, death rate and DALY rate were all positively correlated with the SDI level in 2017 (ρ = 0.21, P = .003;ρ = 0.20, P = .005; ρ = 0.19, P = .006 respectively) 15 . According to the latest estimations, we speculate there should be no de nite linear correlation between the SDI level and the AAPC of mesothelioma burden rates, as the epidemic trend can be in uenced by various factors such as the aging degree, asbestos exposure and regulation, germline predisposition, medical level and data accessibility 25 37 . The current study analyzed the latest data from more areas of 204 countries or territories, standing for more reliable statistical results and less possibility to be affected by particular data.
Most GBD regions hold relatively stable ASRs, or showed a uctuating tendency but eventually going down curves such as Andean Latin America, Central Asia, Western Sub-Saharan Africa and High-income North America having the largest decreasing trends. The exception is Central Europe, Southern Latin America, Oceania and High-income Asia Paci c having obvious increasing trends for age-standardized burden rates, of which governments should be alerted to attach great importance to this matter.
At the national level, Netherlands, Australia, UK have the highest ASIR in 2019, of which UK holds the highest ASMR and ASDR. According to the analysis of the proportion of DALYs due to occupational exposure to asbestos, there are more than 90% of DALYs attributable to occupational exposure in these high-income countries. Moreover, data showed that ASDR attributable to occupational exposure to asbestos in 2019 was positively associated with SDI level, suggesting that the mesothelioma burden increased as the economy developed along with increases in industrial manufacturing and consequent exposure to asbestos 38 . Georgia displayed the highest increasing trend in the ASIR, ASMR and ASDR of mesothelioma during the 30-year period. Croatia, Bahrain, Kuwait and Poland also showed remarkable increase of ASRs. Notably, theses countries also showed increased proportion of DALYs due to occupational exposure to asbestos, with Georgia rose most signi cantly from lower than 50% in 1990 to about 90% in 2019. Besides, the small population of these countries may be another reason for the large variation change. A few more people diagnosed would have a big impact on the overall burden rates data. Nevertheless, these countries should raise more concern of the prevention and management policies for mesothelioma. On the other side, there is a de nite and sometimes substantial fraction of mesotheliomas with no history of asbestos exposure. This fraction has been estimated greater in women and young populations 39 . It is of great signi cance to explore and determine the various contributing factors other than just asbestos in the future.
To our knowledge, this study is the most comprehensive and recent data set available that evaluates the annual incidence, mortality and DALYs associated with mesothelioma and its attributable risk factors over time across a global scale. However, it still has some limitations due to the restrictions of the GBD 2019 database. First, the GBD estimation on mesothelioma is reconstructed by mathematical models based on plenty of data sources with different quality, which may to some extent deviate from the actual data, especially in some underdeveloped regions with extremely scarce prior information, such as Africa and South Asia 16-17 . Second, due to the higher probability of missed diagnosis rate of mesothelioma in developing countries, there is an unavoidable deviation in the estimation of mesothelioma burden. Third, due to the lack of relevant data, we did not estimate the burden of different subtypes of mesothelioma, such as epithelioid, sarcomatoid, and biphasic 40 . Moreover, we only analysed the mesothelioma burden attributed to occupational exposure to asbestos and did not involve an analysis of other potential risk factors, such as other mineral bers and family inheritance 39 . Future research should focus on this aspect, which will help guide different countries and regions to formulate speci c prevention and treatment policies for mesothelioma.
In conclusion, mesothelioma incidence and deaths more than doubled over the study period, with a huge heterogeneity in different locations, sexes and age groups. Mesothelioma is aggressive cancer, often presents in old age, at an advanced stage, and has a poor prognosis. A major risk factor associated with mesothelioma is asbestos exposure. Reducing asbestos exposure through industry regulation with standardized occupational exposure limits remains a key factor to prevent mesothelioma development.
The results of our study can be used by policy makers to allocate resources e ciently for improving the early diagnosis of mesothelioma, reducing its modi able risk factors, and developing novel intervention and treatment strategies to reduce its fatality rate.  icons represent different years. AAPC, average annual percentage change; ASIR, age-standardized incidence rate; SDI, socio-demographic index; GBD, Global Burden of Disease, Injuries, and Risk Factors Study.

Figure 5
The DALYs of mesothelioma attributable to occupational exposure to asbestos: (A) the association between age-standardized DALY rates of mesothelioma attributable to occupational exposure to asbestos in 2019 and SDI in 2019, (B) the proportion of DALYs of mesothelioma attributable to