Background:

In China, the incidence and mortality of prostate cancer are increasing. In this study, we analyzed the spatial-temporal distribution characteristics of prostate cancer incidence and mortality in China and explored the potential associations of socioeconomic, ecological, and meteorologic conditions.

Methods:

Spatial-temporal scan statistics were used to analyze the spatial-temporal patterns of prostate cancer in China from 2012 to 2016. Spatial regression models and the Geodetector method were used to explore the potential associations of anthropogenic and natural factors with prostate cancer.

Results:

The incidence and mortality of prostate cancer in China from 2012 to 2016 rapidly increased. The high incidence and mortality clusters were concentrated in the economically developed Yangtze River Delta region along the southeast coast. Among the 14 selected environmental factors, gross domestic product (GDP) per capita, population density, comprehensive index of environmental pollution discharge, accessibility of health care resources, urbanization rate, and nitrogen dioxide (NO2) had significant positive correlations with prostate cancer incidence and mortality. GDP per capita, urbanization rate, and population density had high explanatory power.

Conclusions:

The high-concentration areas for prostate cancer are located in more economically developed cities. The index of environmental pollution discharge, NO2, and prostate cancer incidence and mortality were positively correlated. The government should advocate increasing the use of clean energy while strengthening the regulation of industrial production to reduce pollutant emissions.

Impact:

To inform the development of prevention and control strategies for prostate cancer in China.

Prostate cancer poses a severe threat to the physical and mental health of the male population and has become a global public health problem. It was the second most common cancer and the fifth leading cause of cancer death among men worldwide in 2020, ranking first in the incidence spectrum of men in 107 countries and first among causes of cancer death among men in 48 countries (1). Trends in prostate cancer incidence and mortality vary widely across countries and regions (2). The incidence of prostate cancer has been increasing globally, with prostate cancer mortality decreasing in developed countries and increasing in developing countries (3). In China, prostate cancer is one of the most common malignancies of the urogenital system. During 2000 to 2016, the incidence and mortality of prostate cancer in China showed increased significantly, with the incidence and mortality rates of 11.05/100,000 and 4.75/100,000, respectively, in 2016, ranking sixth and seventh in the incidence and mortality of malignant tumors in men, respectively (4). According to GLOBOCAN 2020, China accounts for 8.2% and 13.6% of prostate cancer's global incidence and mortality, respectively (5).

The concepts of anthropogenic and natural factors are widely used in environmental science (6, 7). Anthropogenic factors are human production activities and living activities that produce large amounts of exhaust gases, wastewater, and waste, including environmental pollution, and land use, while natural factors include precipitation, temperature, elevation, and others. The factors affecting the spatial and temporal variation of prostate cancer incidence or mortality can also be divided into anthropogenic and natural factors. Anthropogenic factors include economic and social factors [e.g., gross domestic product (GDP) per capita and accessibility of medical resources] and environmental quality factors (e.g., air pollution and land use). Natural factors include geographic factors (e.g., mean elevation and surface relief) and climatic factors (e.g., precipitation and temperature).

Little is known about the etiology of prostate cancer. The known risk factors are advancing age, family history of this malignancy, genetic mutations (e.g., BRCA1 and BRCA2), and conditions (Lynch syndrome; ref. 1). Unfortunately, there is little convincing evidence for lifestyle and environmental factors as a cause of prostate cancer, although smoking, being overweight, and some nutritional factors may increase the risk of developing advanced prostate cancer (8–11). Previous studies on the relationship between prostate cancer and the socioeconomic and ecological environment showed that Western lifestyles, dietary habits, industrial pollution, air quality, and climatic geography might be associated with prostate cancer (12–15). Other studies on the association between natural factors and prostate cancer lack definite evidence, and the results are conflicting (13, 16, 17). Nevertheless, these studies have laid the foundation for further exploration of environmental factors related to prostate cancer. Furthermore, traditional epidemiologic methods cannot explore the geographical characteristics of the disease and its related factors (18). Therefore, we explored the potential associations of physical geography, meteorologic conditions, air quality, and socioeconomic development indicators with prostate cancer using spatial epidemiologic correlation methods.

This article attempts to reveal the potential association between prostate cancer's spatial and temporal distribution characteristics and environmental factors in mainland China using research methods and study contents. We used spatiotemporal scanning statistics (19), combining spatial analysis techniques with traditional epidemiologic research methods (20) to analyze the spatiotemporal distribution characteristics of prostate cancer. We used spatial regression models (21) to identify environmental factors, considering the spatial heterogeneity of the distribution of disease and environmental factors. The Geodetector method (22) was introduced to consider the potential associations of anthropogenic and natural factors such as population density, GDP per capita, environmental quality, urbanization rate, precipitation, and altitude on prostate cancer. Finally, this article proposes countermeasures and recommendations for the prevention and control of prostate cancer.

Data sources and preprocessing

The incidence and mortality data for prostate cancer (ICD-10: C61) were from the National Cancer Center (NCC) using data from cancer registries in China. We collected prostate cancer's municipal-level year-end incidence and mortality data between 2012 and 2016 from the National Cancer Registry Annual Report published by the NCC (23). However, due to insufficient data for most years in several cities, we selected only 224 prefecture-level cities as the study area. Incidence and mortality data were calculated according to the ratio of prostate cancer cases to the number of participants in each municipal unit. The participating population is the total population, without distinction between men and women. There were 20% missing prostate cancer data, which we filled using ordinary kriging interpolation (Supplementary Method S1). We identified several potential environmental correlates associated with prostate cancer from previous studies (16, 17, 24). Regarding anthropogenic factors, economic and social development indicators and air quality were considered. Regarding natural factors, meteorology and physical geography were considered (Supplementary Table S1).

For the economic and social development indicators, GDP per capita (GDPPC), population density (PD), urbanization rate (UR), comprehensive index of environmental pollution discharge (CIEPD), and accessibility of health care resources (AHCR) were included in the study. The data on GDP per capita, PD, and UR of the study area were obtained from the statistical yearbook of each city in China (Supplementary Table S1). The CIEPD was calculated using the entropy weighting method (Supplementary Method S2) for sulfur dioxide emissions from exhaust gases, general industrial solid waste generation, and total wastewater discharge in each city. The AHCR was calculated using the equal-weighted sum method after standardizing the number of health technicians per 1,000 persons, the number of beds in medical institutions per 1,000 persons, and the number of hospitals per 10,000 persons. The data on sulfur dioxide emission from exhaust gases, general industrial solid waste generation, total wastewater discharge, the number of health technicians per 1,000 persons, the number of beds in medical institutions per 1,000 persons, and the number of hospitals per 10,000 persons were derived from the urban statistical yearbook. Air pollutants (PM2.5, PM10, SO2, and NO2) were included (Supplementary Table S1). The PM2.5 concentration values were obtained from the global PM2.5 raster data based on NASA satellite measurements by the Atmospheric Composition Analysis Group at Dalhousie University (https://sites.wustl.edu/acag/datasets/surface-pm2–5/). We used ArcGIS to extract layer information from the raw data and then used zone statistics to match with administrative zones to derive the mean PM2.5 concentrations for each city. PM10, SO2, and NO2 concentration values for each city were obtained from the China Air Quality Online Monitoring and Analysis Platform [https://www.aqistudy.cn/historydata/ (data missing from 2012 to 2013)]. Meteorologic data were provided by the China Meteorological Data Center (http://data.cma.cn/), including monthly temperature, precipitation, and sunshine hours data from meteorologic stations. Monthly data on average temperature, precipitation, and sunshine hours of 613 sites in China from 2012 to 2016 were obtained, and the annual averages of each study area were taken after kriging interpolation (Supplementary Method S1). The calculated average annual temperature (AAT), average annual precipitation (AAP), and average annual sunshine hours (AASH) were included in this study (Supplementary Table S1). Among the physical geography indicators, the relief degree of land surface (RDLS) and average elevation (AE) of each city were included (Supplementary Table S1). The topographic elevation and relief degree of land surface distributions were obtained from Advanced Spaceborne Thermal Emission and Reflection Radiometer Global Digital Elevation Model (30 m spatial resolution; http://datamirror.csdb.cn/index.jsp/), and the RDLS and AE for each city was calculated on the basis of the city's map.

Research framework

This study was conducted in the following steps (Supplementary Fig. S1). First, five datasets were collected and preprocessed separately. Then, spatiotemporal disease data analysis was performed to detect spatial heterogeneity of their distributions. Next, ordinary least squares (OLS) model, spatial lag model (SLM), and spatial error model (SEM) were compared to perform a single-factor regression analysis of potential environmental factors with prostate cancer incidence and mortality with suitable models to demonstrate whether explanatory variables were correlated with prostate cancer incidence and mortality. Finally, the relevant factors were included in the geographic detector model for analysis to quantify the degree of association of the explanatory variables on prostate cancer incidence and mortality and the interaction of the explanatory variables.

Spatiotemporal scan statistics

We performed a retrospective spatiotemporal scan statistical analysis (19) of the discrete Poisson distribution model for prostate cancer incidence and mortality in China from 2012 to 2016 in SaTScan. The method is based on a moving circular active window. By dynamically changing the center's position and the window's radius, the difference in prostate cancer incidence and mortality between inside and outside the window after each movement is compared according to the chosen model. The log-likelihood ratio (LLR) and the corresponding P value are calculated for different windows, and the relative risk (RR) is also calculated. The P values of LLR were simulated using the Monte Carlo method. P < 0.05 and a larger LLR value indicates a more significant proportion of the risk of prostate cancer incidence and death in the scan window, which is a more likely area of aggregation. To explore prostate cancer's incidence and mortality characteristics in different locations and times, the maximum spatial radius was set up to 50% of the population at risk, and the maximum temporal height (time range of clusters) was set to 50% of the study period. The LLR expression is:
formula
where n is the number of prostate cancer incidence and death cases in the scan window, N is the total number of prostate cancer incidence and death cases in the population, and E is the expected number of prostate cancer incidence and death cases under the null hypothesis. If the null hypothesis is true, I = 1 when the scan window has more prostate cancer incidence and death cases than expected; otherwise, it is 0.

Spatial regression model

Spatial regression analysis of environmental influences on prostate cancer in China was performed using GeoDa software, followed by identifying associated factors. The OLS model does not consider the autocorrelation and heterogeneity of spatial data. We, therefore, adopted two other spatially constant coefficient models, SLM and SEM, as spatial extensions of the conventional global regression model, OLS (21, 25). The expression of the SLM is as follows (21, 25, 26):
formula
where ρ is an autoregressive parameter expressing the spatial dependence in the lattice, the spatial lag is a smoothing term of the weighted average of adjacent values, and the error ϵ follows a normal distribution.
The expression of the SEM is as follows (21, 25, 26):
formula
where y is the N × 1 vector of the dependent variable and x denotes the N × K matrix of the explanatory variables; β is the K × 1 vector of parameters; u is the N × 1 vector of residuals; the scalar λ is the spatial autocorrelation coefficient of errors; W denotes the N × N spatial weight matrix containing contiguity or distance functions; ϵ is the normal distribution error vector.

In the model selection of SLM and SEM, Lagrange multipliers (LM-lag and LM-error) and their robust LM diagnostics (RLM-lag and RLM-error) are used to determine which model is more appropriate. Anselin proposed a discriminant for which model is more appropriate (27, 28), that is, if the statistical LM-lag is more significant than the LM-error and the robust LM-lag is significant while the robust LM-error is not, then the SLM model is appropriate; otherwise, the SEM model is appropriate (27, 28).

Geodetector method

Geodetector was used to assess the association between prostate cancer and socioecological factors; it is based on the principle that if a potential factor is associated with a disease, that factor may show a similar spatial and temporal distribution to that disease. Geodetector uses four detectors to assess the environmental risk to health based on spatial variation analysis of geographic strata: factor detection, interaction detection, risk detection, and ecological detection. Factor detection, measured by q values (29):
formula
where N is the regional population and h is the stratification of the influencing factors, |${N}_h$|and N represent the number of spatial units in layer h and the entire study area, respectively; |$\sigma _h^2$| and |${\sigma }^2$| represent the variance of Y values in layer h and the entire study area, respectively. q has a value range of [0, 1], with larger values indicating more substantial explanatory power of the independent variable X for attribute Y.

The interaction detector can also be used to study the interaction between any two factors (represented by ∩). If q(X1∩X2) > q(X1) or q(X2) indicates enhancement, if q(X1∩X2) > q(X1) + q(X2) indicates nonlinear enhancement, and if q(X1∩X2) > q(X1) and q(X2) indicates bivariate enhancement, If q(X1∩X2) < q(X1) + q(X2) indicates weakening, if q(X1∩X2) < q(X1) and q(X2) indicates nonlinear weakening, if q(X1∩X2) < q(X1) or q(X2) indicates univariate weakening, If q(X1∩X2) = q(X1) + q(X2) represents independence (29).

Data availability

The data were drawn partly from public domain sources, as detailed in Supplementary Table S1. The data generated in this study are available upon request from the corresponding author.

Descriptive statistics of prostate cancer

In the study area, there were 76,972 prostate cancer cases, and 33,344 deaths were reported from 2012 to 2016, for a total incidence rate of 4.95/100,000 and a total mortality rate of 2.14/100,000 (Supplementary Table S2). There was a linear increase in the incidence and mortality of prostate cancer in China from 2012 to 2016, reaching 33.89% and 34.97%, respectively. The spatial distribution of the average incidence and mortality rate of prostate cancer in various cities from 2012 to 2016 is shown in Fig. 1.

Spatial-temporal distribution characteristics of prostate cancer

Analysis of the SaTScan spatiotemporal scan of prostate cancer incidence in China from 2012 to 2016 showed that nine high-incidence clusters were identified in the study area (Fig. 2A). The clusters were sorted in descending by LLR (Table 1). Theoretically, the window with the maximum likelihood value is the most likely cluster, which is least likely to be due to chance. The first-level cluster with high incidence (LLR = 4611.19, RR = 3.05, P < 0.001) was located along the southeast coast and included 15 cities with a radius of 262.55 km, with 11,320 observed cases and 4,119 expected cases, clustered from 2015 to 2016 (Table 1). The secondary cluster (LLR = 664.64, RR = 2.18, P < 0.001) was Beijing, where 2,850 observed cases and 1,332 expected cases were clustered from 2015 to 2016 (Table 1). The tertiary cluster (LLR = 145.16, RR = 1.37, P < 0.001) was located in Guangdong and Guangxi, covering 14 cities with a radius of 270.75 km, in which 3,375 observed cases and 2,494 expected cases were clustered from 2015 to 2016 (Table 1). The fourth-level cluster was Wuhan and Xiaogan, and the fifth- to ninth-level clusters were Kunming, Urumqi, Chongqing, Chengdu, and Lanzhou, respectively (Table 1; Fig. 2A).

Nine high mortality clusters were also identified in the spatiotemporal scan of prostate cancer mortality (Fig. 2B). Similar to the high incidence clusters, the first-level cluster (LLR = 1260.50, RR = 2.91, P < 0.001) was located on the southeast coast, including seven cities with a radius of 150.88 km, with 3,229 observed cases and 1,184 expected cases, clustered from 2015 to 2016 (Table 2). The secondary cluster (LLR = 312.97, RR = 1.91, P < 0.001) was Beijing and Tianjin, with a radius of 113.67 km, including 1,900 observed cases and 1,021 expected cases, with an aggregation time of 2015 to 2016 (Table 2). The tertiary cluster (LLR = 82.35.16, RR = 1.59, P < 0.001) was mainly located in Guangdong, covering four cities with a radius of 62.64 km, including 909 observed cases and 578 expected cases, with an aggregation time of 2015 to 2016 (Table 2). The fourth- to ninth-level clusters covered Kunming, Wuhan, Xiaogan, Chengdu, Lanzhou, Guilin, Liuzhou, and Changsha (Table 2; Fig. 2B).

Environmental factors associated with the spatial and temporal distribution of prostate cancer

Selection of evaluation indicators

The spatial distribution of the mean of all anthropogenic and natural factors is shown in Supplementary Fig. S2. The potential factors were screened by the regression model. First, OLS was used to perform a single-factor regression analysis of the environmental factors and prostate cancer incidence and mortality, respectively, and Moran I index of residuals was obtained, all of which were spatially correlated. LM and RLM were analyzed, and suitable spatial regression models were fitted according to model selection criteria. Table 3 displays the results of the spatial regression analysis between the 5-year mean incidence of prostate cancer and each factor. The fitted results showed that GDPPC, PD, CIEPD, UR, AHCR, and NO2 were significantly positively associated with prostate cancer incidence. AE was negatively associated with prostate cancer incidence. Table 4 displays the results of spatial regression analysis between the mean 5-year prostate cancer mortality rate and each spatial factor. GDPPC, PD, CIEPD, UR, AHCR, NO2, and AAP were positively correlated with prostate cancer mortality; AE was negatively correlated with prostate cancer mortality. Because the remaining factors were not significantly correlated with prostate cancer incidence and mortality at the 0.05 level, these factors were excluded from the Geodetector method.

Analysis of environmental factors

The average incidence of prostate cancer from 2012 to 2016 was used as the dependent variable, and the screened anthropogenic and natural factors were used as independent variables. The natural breakpoint method (30) was used to stratify the independent variables. The numerical quantities were transformed into type quantities for factor detection to derive the degree of association of each variable with prostate cancer incidence and the degree of interaction between its bivariate variables (Supplementary Table S3). The q-values on the diagonal represent the explanatory power of individual factors, that is, the independent effect of every single factor on prostate cancer incidence, and the other values indicate the interaction between two factors on prostate cancer incidence. The factor detector showed that the order of q-value magnitude of each factor was GDPPC (0.31) > UR (0.29) > PD (0.26) > AHCR (0.18) > NO2 (0.13) > CIEPD (0.12) > AE (0.09), all with P values less than 0.001 (Supplementary Table S3). The association between GDPPC and prostate cancer incidence showed the strongest positive correlation (Table 3; Supplementary Table S3). The interaction detector showed that the interaction between any two factors was enhanced relative to their independent effects (Supplementary Table S3). The q-values of GDPPC ∩ UR, GDPPC ∩ AHCR, and UR ∩ AHCR were greater than the q-values of the single factors, showing a bivariate enhancement, indicating that the interaction of the detection factors had a significant effect on the incidence of prostate cancer. The q-values of any two remaining factor interactions were greater than the sum of the q-values of the two factors, showing a nonlinear enhancement, indicating that their interactions were all greater than the sum of their independent effects. The combined effect of GDPPC and PD had the most significant association with prostate cancer incidence, with an explanatory power of 0.63 (Supplementary Table S3). High GDPPC and high PD were associated with a high incidence of prostate cancer (Table 3; Supplementary Table S3).

Similarly, when mean prostate cancer mortality was used as the dependent variable, the results of the factor detector showed q-values in the order of UR (0.27) > GDPPC (0.26) > PD (0.25) > AHCR (0.17) > AAP (0.15) > AE (0.10) > NO2 (0.09) > CIEPD (0.08), all with P values less than 0.001 (Supplementary Table S4). UR showed the strongest positive correlation for prostate cancer mortality (Table 4; Supplementary Table S4). The results of the interaction detector showed that the q-values of GDPPC ∩ UR showed a bivariate enhancement, and the interactions of the remaining factors were nonlinearly enhanced (Supplementary Table S4). The interaction of AHCR and PD had the most substantial explanatory power of 0.68 for prostate cancer mortality (Supplementary Table S4). High AHCR and high PD were associated with high prostate cancer mortality (Table 4; Supplementary Table S4).

The main finding was that within the study area, the incidence and mortality of prostate cancer in China increased from 2012 to 2016, consistent with other studies (31, 32). The incidence of prostate cancer in China is relatively low; however, it is the fastest-growing common cancer in Chinese men (33). In contrast to developed areas such as Europe and the United States, where the incidence of prostate cancer has plateaued or decreased (1, 2), China is a developing country experiencing rapid changes in social and economic development, with an aging population, westernized lifestyles, and changes in diet; the incidence and mortality of prostate cancer in China is rapidly increasing.

Significant geographical differences exist in the incidence and mortality of prostate cancer in China (Fig. 1). Differences in environmental factors in different regions and prostate cancer screening strategies may be partly responsible for the heterogeneity in the spatial distribution of prostate cancer. Previous studies on prostate cancer epidemiology in China focused on comparing urban-rural differences (31, 32), with few studies specific to administrative divisions. In the United States, there are significant geographical differences in the incidence of prostate cancer (34), caused by ethnicity and genetic predisposition (35), and by regional differences in PSA testing (36).

Differences in the spatial distribution of prostate cancer were reflected in the results of the spatiotemporal scans. In the survey of prostate cancer incidence, we found that the first clusters of high incidence were concentrated in Shanghai, southern Jiangsu, and northern Zhejiang (Table 1; Fig. 1A), which are in the coastal zone of eastern China and located in the Yangtze River Delta economic development circle. The relatively good socioeconomic development in these areas may be partly responsible for the high incidence of clustering. Because the promotion of health security systems accompanies socioeconomic development, the prevalence and refinement of technologies such as pathologic tissue biopsy, MRI, and biomarker testing have increased the rate of early diagnosis of prostate cancer (37). Furthermore, areas with better economic development may have more westernized lifestyles with increased opportunities to consume red meat, fat, dairy products, fried foods, and eggs. Previous studies demonstrated that these personal dietary habits might increase the risk of prostate cancer in men (38–40). In the spatiotemporal survey of prostate cancer mortality, the high mortality aggregates mostly overlap with high incidence rates (Fig. 1). These findings suggest that the higher incidence rate may be the primary reason for these clusters’ higher prostate cancer mortality. This is because the prognosis of patients with prostate cancer in China is poor, less than the global average, and much lower than that of developed countries such as Europe and the United States (41).

The spatial regression analysis showed that among the anthropogenic factors, GDPPC, PD, CIEPD, UR, AHCR, and NO2 were positively associated with the incidence and mortality of prostate cancer. The accelerated urbanization in China in recent years will undoubtedly bring about inevitable public health problems (42). Under rapid urbanization, not only prostate cancer but also thyroid cancer and hand, foot, and mouth disease positively correlate with socioeconomic factors (42, 43). The CIEPD is a certain degree of representation of the degree of industrial pollution in a city. Therefore, areas with high industrial pollution have a relatively higher incidence and mortality of prostate cancer, as demonstrated by a study in Shanghai (24). In Western countries, the high incidence of prostate cancer is closely related to the widespread availability of PSA testing (44). It has become well established that the high sensitivity of PSA screening leads to a significant increase in prostate cancer (36). In part, this variation in incidence rates across populations can be attributed to differences in diagnostic intensity arising from PSA screening (11). Studies in China have also suggested that PSA screening may be responsible for the increased prostate cancer incidence in recent years (45). In contrast, better AHCR may increase access to physical screening, which may be associated with higher incidence rates. Most patients with prostate cancer in China are characterized by advanced stage at the time of diagnosis, with poor prognosis and high incidence accompanied by high mortality (46); therefore, the AHCR may be associated with mortality. Previous studies have been controversial and inconclusive regarding whether air pollutants affect prostate cancer. Studies in Saudi Arabia and Canada showed an increased risk of prostate cancer in men exposed to high levels of NO2 (13, 47). A study in Saxony, Germany, showed that NO2 and PM10 are moderately associated with prostate cancer (16), but a study in Spain showed that PM2.5 and NO2 do not seem to be associated with the risk of prostate cancer (17). Among natural factors, AAP was positively associated with prostate cancer mortality, and AE was negatively correlated with prostate cancer incidence and mortality. Areas with more rainfall will have relatively less exposure to sunlight. A Spanish ecology-type study assessed whether climatic factors (temperature, rainfall, and the number of sunlight hours per year) might influence prostate cancer-related mortality over a 5-year period (48). It was concluded that prostate cancer mortality was significantly higher in areas with less exposure to sunlight than in other areas. Occupational exposure to solar UV radiation and the risk of prostate cancer was studied in a population-based case–control study in Canada (49). It was concluded that the disease risk was reduced among workers with the most prolonged and intense sun exposure. Precipitation may not be a direct cause of increased cancer incidence but may act as a “vector” for predisposing factors, increasing an individual's risk of exposure to carcinogens or increasing the likelihood of carcinogenic substances being produced by nature (50). The mechanisms by which precipitation affects cancer are unclear, and further studies are needed to confirm this. An earlier study suggested that the lower cancer incidence at higher elevations was due to enhanced daylight-induced vitamin D production (51). A study in the United States concluded that higher elevation counties had lower cancer mortality among individual races compared with lower elevation counties, suggesting the presence of radiation hormesis (52).

In the current study, combined with Geodetector's analysis, anthropogenic factors had greater explanatory power for prostate cancer. The explanatory power of the relevant factors for prostate cancer is greatly enhanced when the interaction is considered. GDPPC had the most significant explanatory power for the distribution of prostate cancer incidence. The interaction of GDPPC and PD had the most significant power in determining the incidence of prostate cancer. The incidence of prostate cancer was higher in areas with high GDPPC and PD. The UR has the most significant explanatory power in analyzing factors associated with prostate cancer mortality. The interaction between AHCR and PD had the most significant determinants of prostate cancer mortality. Areas with high AHCR and high PD had higher prostate cancer mortality.

This study has several limitations. First, we used publicly available prostate cancer statistics from 2012 to 2016 from the NCC of China, with limited coverage of tumor surveillance sites, and some areas were excluded considering the continuity and availability of data. Second, socioecological factors are macroscopic, and the potential association with prostate cancer is not evident; therefore, it is impossible to determine whether the factors have a direct or indirect impact. Third, this is an ecological study; therefore, no inferences can be drawn regarding individual-level patterns of environmental risk and prostate cancer. The results of this study should be interpreted with caution, and further field studies or individual-level studies are needed to improve them. Finally, cancer is a chronic disease caused by several factors and a latency period; the lag of various factors affecting the disease was not considered in the study. In this article, only 5 years of data were selected for analysis, and a more extended time series study was not conducted. In a subsequent study, we will broaden the time and space to provide a detailed theoretical basis for prostate cancer studies.

This article combines spatial analysis techniques with traditional epidemiologic analysis methods to analyze the spatial-temporal distribution characteristics of prostate cancer incidence and mortality in mainland China during 2012 to 2016 and a comprehensive quantitative study of the potential association between anthropogenic and natural factors. From a temporal perspective, the incidence and mortality of prostate cancer showed an increasing trend. The high incidence and mortality clusters were found in the Yangtze River Delta region of the southeast coast. Among the anthropogenic factors, economic and social development indicators, NO2, and prostate cancer incidence and mortality all showed positive correlations, among which the CIEPD and NO2 concentration should be studied. Among the natural factors, the AE was significantly and negatively correlated with prostate cancer incidence and mortality, and AAP was positively associated with prostate cancer mortality. GDPPC, UR, and PD have substantial power to determine the distribution of prostate cancer incidence and mortality. Therefore, a comprehensive prevention and control strategy needs to be developed on the basis of the distribution characteristics of prostate cancer combined with key anthropogenic and natural factors tailored to local conditions.

No disclosures were reported.

M. Zhang: Software, formal analysis, methodology, writing–original draft. X. Dai: Formal analysis, validation, investigation. G. Chen: Formal analysis, investigation, visualization. Y. Liu: Resources, software, methodology. Z. Wu: Resources, investigation, methodology. C. Ding: Resources, investigation. Y. Chang: Resources, investigation. H. Huang: Data curation, methodology, writing–original draft, writing–review and editing.

This study was financially supported by the Science and Technology Planning Project of Wenzhou (S2020002), and the Research Project of Wenzhou Medical University (KJHX2014). H. Huang received financial support for the above project.

The publication costs of this article were defrayed in part by the payment of publication fees. Therefore, and solely to indicate this fact, this article is hereby marked “advertisement” in accordance with 18 USC section 1734.

Note: Supplementary data for this article are available at Cancer Epidemiology, Biomarkers & Prevention Online (http://cebp.aacrjournals.org/).

1.
Sung
H
,
Ferlay
J
,
Siegel
RL
,
Laversanne
M
,
Soerjomataram
I
,
Jemal
A
, et al
.
Global Cancer Statistics 2020: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries
.
CA Cancer J Clin
2021
;
71
:
209
49
.
2.
Culp
MB
,
Soerjomataram
I
,
Efstathiou
JA
,
Bray
F
,
Jemal
A
.
Recent global patterns in prostate cancer incidence and mortality rates
.
Eur Urol
2020
;
77
:
38
52
.
3.
Zhai
Z
,
Zheng
Y
,
Li
N
,
Deng
Y
,
Zhou
L
,
Tian
T
, et al
.
Incidence and disease burden of prostate cancer from 1990 to 2017: results from the Global Burden of Disease Study 2017
.
Cancer
2020
;
126
:
1969
78
.
4.
Zheng
R
,
Zhang
S
,
Zeng
H
,
Wang
S
,
Sun
K
,
Chen
R
, et al
.
Cancer incidence and mortality in China, 2016
.
J Natl Cancer Cent
2022
;
2
:
1
9
.
5.
Cao
W
,
Chen
HD
,
Yu
YW
,
Li
N
,
Chen
WQ
.
Changing profiles of cancer burden worldwide and in China: a secondary analysis of the global cancer statistics 2020
.
Chin Med J
2021
;
134
:
783
91
.
6.
Huang
H
,
Chen
D
,
Zhang
B
,
Zeng
L
,
Dahlgren
R
.
Modeling and forecasting riverine dissolved inorganic nitrogen export using anthropogenic nitrogen inputs, hydroclimate, and land-use change
.
J Hydrol
2014
;
517
:
95
104
.
7.
Huang
S
,
Xiao
L
,
Zhang
Y
,
Wang
L
,
Tang
L
.
Interactive effects of natural and anthropogenic factors on heterogenetic accumulations of heavy metals in surface soils through geodetector analysis
.
Sci Total Environ
2021
;
789
:
147937
.
8.
Islami
F
,
Moreira
DM
,
Boffetta
P
,
Freedland
SJ
.
A systematic review and meta-analysis of tobacco use and prostate cancer mortality and incidence in prospective cohort studies
.
Eur Urol
2014
;
66
:
1054
64
.
9.
Kelly
SP
,
Graubard
BI
,
Andreotti
G
,
Younes
N
,
Cleary
SD
,
Cook
MB
.
Prediagnostic body mass index trajectories in relation to prostate cancer incidence and mortality in the PLCO Cancer Screening Trial
.
J Natl Cancer Inst
2016
;
109
:
djw225
.
10.
Xu
C
,
Han
FF
,
Zeng
XT
,
Liu
TZ
,
Li
S
,
Gao
ZY
.
Fat intake is not linked to prostate cancer: a systematic review and dose-response meta-analysis
.
PLoS One
2015
;
10
:
e0131747
.
11.
Pernar
CH
,
Ebot
EM
,
Wilson
KM
,
Mucci
LA
.
The epidemiology of prostate cancer
.
Cold Spring Harb Perspect Med
2018
;
8
:
a030361
.
12.
Lu
W
,
Chen
H
,
Niu
Y
,
Wu
H
,
Xia
D
,
Wu
Y
.
Dairy products intake and cancer mortality risk: a meta-analysis of 11 population-based cohort studies
.
Nutr J
2016
;
15
:
91
.
13.
Parent
ME
,
Goldberg
MS
,
Crouse
DL
,
Ross
NA
,
Chen
H
,
Valois
MF
, et al
.
Traffic-related air pollution and prostate cancer risk: a case-control study in Montreal, Canada
.
Occup Environ Med
2013
;
70
:
511
8
.
14.
St-Hilaire
S
,
Mannel
S
,
Commendador
A
,
Mandal
R
,
Derryberry
D
.
Correlations between meteorological parameters and prostate cancer
.
Int J Health Geogr
2010
;
9
:
19
.
15.
Su
SY
,
Liaw
YP
,
Jhuang
JR
,
Hsu
SY
,
Chiang
CJ
,
Yang
YW
, et al
.
Associations between ambient air pollution and cancer incidence in Taiwan: an ecological study of geographical variations
.
BMC Public Health
2019
;
19
:
1496
.
16.
Datzmann
T
,
Markevych
I
,
Trautmann
F
,
Heinrich
J
,
Schmitt
J
,
Tesch
F
.
Outdoor air pollution, green space, and cancer incidence in Saxony: a semi-individual cohort study
.
BMC Public Health
2018
;
18
:
715
.
17.
O'Callaghan-Gordo
C
,
Garcia
B
,
Kogevinas
M
,
Fernandez-Tardon
G
,
Dierssen-Sotos
T
,
Jiménez
J
, et al
.
Air pollution and prostate cancer risk in the multicase-control study in Spain (MCC-Spain)
.
In: ISEE Conference Abstracts
;
2018
.
2017. Abstract nr 518
.
18.
Mahara
G
,
Wang
C
,
Yang
K
,
Chen
S
,
Guo
J
,
Gao
Q
, et al
.
The association between environmental factors and scarlet fever incidence in Beijing region: using GIS and spatial regression models
.
Int J Environ Res Public Health
2016
;
13
:
1083
.
19.
Kulldorff
M
.
Prospective time periodic geographical disease surveillance using a scan statistic
.
J R Stat Soc Ser A
2001
;
164
:
61
72
.
20.
Elliott
P
,
Wartenberg
D
.
Spatial epidemiology: current approaches and future challenges
.
Environ Health Perspect
2004
;
112
:
998
1006
.
21.
Anselin
L
.
Spatial econometrics: methods and models
.
Dordrecht, the Netherlands
:
Kluwer Academic Publishers
;
1988
.
22.
Wang
JF
,
Xu
CD
.
Geodetector: principle and prospective
.
Acta Geographica Sinica
2017
;
72
:
116
34
.
23.
He
J
.
Chinese cancer registry annual report 2019
.
Beijing
;
2021
.
24.
Cong
X
.
Air pollution from industrial waste gas emissions is associated with cancer incidences in Shanghai, China.
Environ Sci Pollut Res Int
2018
;
25
:
13067
78
.
25.
Kissling
WD
,
Carl
G
.
Spatial autocorrelation and the selection of simulta-neous autoregressive models
.
Global Ecol Biogeogr
2007
;
17
:
59
71
.
26.
Li
F
,
Wang
X
,
Liu
H
,
Li
X
,
Zhang
X
,
Sun
Y
, et al
.
Does economic development improve urban greening? Evidence from 289 cities in China using spatial regression models
.
Environ Monit Assess
2018
;
190
:
541
.
27.
Anselin
L
.
Spatial externalities, spatial multipliers, and spatial econometrics
.
Int Reg Sci Rev
2003
;
26
:
153
66
.
28.
Anselin
L
,
Florax
R
.
Introduction
.
In:
Anselin
L
,
Florax
R
,
editors.
New directions in spatial econometrics
.
Berlin
:
Springer-Verlag
;
1995
. p.
3
18
.
29.
Wang
JF
,
Li
XH
,
Christakos
G
,
Liao
YL
,
Zhang
T
,
Gu
X
, et al
.
Geographical detectors-based health risk assessment and its application in the neural tube defects study of the Heshun region, China
.
Int J Geogr Inf Sci
2010
;
24
:
107
27
.
30.
Cao
F
,
Ge
Y
,
Wang
JF
.
Optimal discretization for geographical detectors-based risk assessment
.
Mapp Sci Rem Sens
2013
;
50
:
78
92
.
31.
Gu
X
,
Zheng
R
,
Zhang
S
,
Zeng
H
,
Sun
K
,
Zou
X
, et al
.
Analysis on trend of prostate cancer incidence and change in cancer registration areas of China, 2000 to 2014
.
Chin J Prev Med
2018
;
52
:
586
92
.
32.
Han
S
,
Liu
F
,
Xing
N
.
Trend analysis of the incidence of prostate cancer in cancer registration areas of China between 1988 and 2015
.
China J Urol
2021
;
43
:
51
5
.
33.
Chen
W
,
Zheng
R
,
Baade
PD
,
Zhang
S
,
Zeng
H
,
Bray
F
, et al
.
Cancer statistics in China, 2015
.
CA Cancer J Clin
2016
;
66
:
115
32
.
34.
Scott
LC
,
Kuo
TM
,
Il'yasova
D
,
Mobley
LR
.
Geospatial analysis of multiple cancers in individuals in the US, 2004–2014
.
Ann Cancer Epidemiol
2021
;
5
:
2
.
35.
Rebbeck
TR
,
Devesa
SS
,
Chang
BL
,
Bunker
CH
,
Cheng
I
,
Cooney
K
, et al
.
Global patterns of prostate cancer incidence, aggressiveness, and mortality in men of African descent
.
Prostate Cancer
2013
;
2013
:
560857
.
36.
Kensler
KH
,
Rebbeck
TR
.
Cancer progress and priorities: prostate cancer
.
Cancer Epidemiol Biomarkers Prev
2020
;
29
:
267
77
.
37.
McGuire
S
.
World Cancer Report 2014. Geneva, Switzerland: World Health Organization, International Agency for Research on Cancer, WHO Press, 2015
.
Adv Nutr
2016
;
7
:
418
9
.
38.
Gandaglia
G
,
Leni
R
,
Bray
F
,
Fleshner
N
,
Freedland
SJ
,
Kibel
A
, et al
.
Epidemiology and prevention of prostate cancer
.
Eur Urol Oncol
2021
;
4
:
877
92
.
39.
López-Plaza
B
,
Bermejo
LM
,
Santurino
C
,
Cavero-Redondo
I
,
Álvarez-Bueno
C
,
Gómez-Candela
C
.
Milk and dairy product consumption and prostate cancer risk and mortality: an overview of systematic reviews and meta-analyses
.
Adv Nutr
2019
;
10
:
S212
23
.
40.
Lippi
G
,
Mattiuzzi
C
.
Fried food and prostate cancer risk: systematic review and meta-analysis
.
Int J Food Sci Nutr
2015
;
66
:
587
9
.
41.
Fu
ZT
,
Guo
XL
,
Zhang
SW
,
Zheng
RS
,
Zeng
HM
,
Chen
R
, et al
.
Statistical analysis of incidence and mortality of prostate cancer in China, 2015
.
Chinese J Oncol
2020
;
42
:
718
22
.
42.
Xu
C
,
Zhang
X
,
Xiao
G
.
Spatiotemporal decomposition and risk determinants of hand, foot and mouth disease in Henan, China
.
Sci Total Environ
2019
;
657
:
509
16
.
43.
Fei
X
,
Chen
W
,
Zhang
S
,
Liu
Q
,
Zhang
Z
,
Pei
Q
.
The spatio-temporal distribution and risk factors of thyroid cancer during rapid urbanization-a case study in China
.
Sci Total Environ
2018
;
630
:
1436
45
.
44.
Jahn
JL
,
Giovannucci
EL
,
Stampfer
MJ
.
The high prevalence of undiagnosed prostate cancer at autopsy: implications for epidemiology and treatment of prostate cancer in the prostate-specific Antigen-era
.
Int J Cancer
2015
;
137
:
2795
802
.
45.
Liu
X
,
Yu
C
,
Bi
Y
,
Zhang
ZJ
.
Trends and age-period-cohort effect on incidence and mortality of prostate cancer from 1990 to 2017 in China
.
Public Health
2019
;
172
:
70
80
.
46.
Liu
J
,
Dong
YC
,
Xu
DB
,
Yang
H
,
Ding
HS
,
Zhou
X
, et al
.
Research status on the factors affecting prognosis in patients with prostate cancer
.
J Pract Med
2021
;
37
:
1365
70
.
47.
Al-Ahmadi
K
,
Al-Zahrani
A
.
NO2 and cancer incidence in Saudi Arabia
.
Int J Environ Res Public Health
2013
;
10
:
5844
62
.
48.
Santos Arrontes
D
,
García González
JI
,
Martín Muñoz
MP
,
Castro Pita
M
,
Mañas Pelillo
A
,
Paniagua Andre´s
P
.
[Will the climate change affect the mortality from prostate cancer?]
Arch Esp Urol
2007
;
60
:
119
.
49.
Peters
CE
,
Demers
PA
,
Kalia
S
,
Hystad
P
,
Villeneuve
PJ
,
Nicol
AM
, et al
.
Occupational exposure to solar ultraviolet radiation and the risk of prostate cancer
.
Occup Environ Med
2016
;
73
:
742
8
.
50.
Shah
V
,
Rieger
RH
,
Pan
LX
.
Precipitation and climate zone explains the geographical disparity in the invasive cancer incidence rates in the United States
.
Environ Eng Sci
2019
;
36
:
1452
8
.
51.
Hayes
DP
.
Cancer protection related to solar ultraviolet radiation, altitude and vitamin D
.
Med Hypotheses
2010
;
75
:
378
82
.
52.
Hart
J
.
Cancer mortality for a single race in low versus high elevation counties in the u.s
.
Dose Response
2011
;
9
:
348
55
.