Spatial co-occurrence of multiple cancers: a population-based study using national cancer registry data in China

doi:10.21203/rs.3.rs-3123653/v1

Download PDF

Research Article

Spatial co-occurrence of multiple cancers: a population-based study using national cancer registry data in China

https://doi.org/10.21203/rs.3.rs-3123653/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

Background

Showing the spatial co-occurrence of multiple types of cancers supports geographical targeting and prioritization, because different cancer types often share common causes. However, the variation in incidence between different cancer types and regional differences within each cancer pose a major challenge to etiological research. This study aimed to assess the spatial co-occurrence of multiple cancers in mainland China to accelerate the identification of causes and development of tailored prevention policies.

Methods

We obtained cancer incidence data for 2016 for 13 cancers from the China Cancer Registry Annual Report. We proposed a design framework to assess the spatial co-occurrence of multiple cancer types using Moran’s I, and identified the level of risk of co-occurrence by area. We used negative binomial regression to obtain the incidence rate ratio for three risk-level areas, and the population attributable fraction and expected excess cases to estimate the proportion of cancer incidence attributable in different risk areas.

Results

The high-risk areas (17.1%) for cancer co-occurrence were mainly in the east and northeast, the low-risk areas (30.7%) were mainly in the south, and medium-risk areas (52.2%) were evenly distributed throughout China. The incidence rate ratio (95% confidence interval) was 1.61 (1.53–1.69) for high-risk areas, and 1.19 (1.14–1.23) for medium-risk areas compared to low-risk areas. Most provinces had two to three grades of risk areas. Zhejiang had the highest proportion of high-risk areas (85.7%), and Hunan the highest proportion of low-risk areas (86.4%). The most frequently observed cancer co-occurrence patterns were lung and pancreatic in the 15 provinces with high-risk areas; leukemia, brain tumor, bone, and gallbladder cancer in the 30 provinces with medium-risk areas; and pancreatic, lung and stomach cancer in the 24 provinces with low-risk areas. The population attributable factor ranged from 7.6–37.9% for high-risk areas and 1.8–15.9% for medium-risk areas. There were 44,568 expected excess cases in all areas. The highest number of expected excess cases was in Jiangsu and the lowest in Tibet.

Conclusion

This research framework could help to locate areas at high risk of cancer co-occurrence and identify common risk factors, guiding the development of tailored prevention policies.

spatial co-occurrence

multiple cancers

incidence rate ratio

high-risk areas

cancer co-occurrence pattern

Cancer is the second most common cause of death worldwide and is a huge burden on healthcare systems and economies. [1] There are more than 200 types of cancer. When classified according to their site of origin in the body, 36 types of cancer provided up to 19 million new cases in 185 countries in 2020. [2] This number is expected to increase to 34 million by 2070 because of population aging and growth. [3] The wide diversity of cancers poses a major challenge to etiological research using conventional study designs and analysis. However, the rapidly increasing incidence adds to the urgency of developing effective measures for cancer prevention and control.

The development of all cancers has a common biological process of the genetic transformation of normal cells to highly malignant derivatives.[4] There are numerous common factors (infectious, behavioral, and environmental) that influence this multistep process, [5, 6] suggesting that the occurrence of different types of cancer may be interrelated. Studies at the individual level have shown that two or more cancers of different origins may occur in the same patient. [7] Spatial analysis can help to clarify the phenomenon of co-occurrence of multiple cancers at the population level. For example, population-based cancer registry data, which is now readily available in many countries around the world, has shown that stomach and esophageal cancers are interrelated and more common in East Asia; similar findings have been observed with brain cancer and lymphoma in Southern Europe, and kidney cancer and leukemia in North America. [8] Big data platforms such as these provide an overall picture of the current burden, geographical distribution, and even temporal trends in different cancers. However, the patterns of cancer co-occurrence remain less comprehensively described.

A better understanding of cancer co-occurrence would provide a unique opportunity for geographic targeting and prioritization, which is important for resource-constrained countries such as China. It would also facilitate the search for causes (internal or external) that might explain why populations living in different areas are at higher risk for different cancers. However, analyses of multiple cancer types are methodologically challenging because it is tempting to treat cancers unequally (because of differences in the magnitude of incidence) and because there are large differences in spatial patterns within and between cancer types.

A better understanding of cancer co-occurrence would provide opportunities for geographical targeting and prioritization, which is important for resource-constrained countries such as China. It would also facilitate the search for causes (internal or external) that might explain why populations living in certain areas are at higher risk of particular cancers. However, analyses of multiple cancer types are methodologically challenging because it is tempting to treat cancers unequally (because of differences in the magnitude of incidence) and because there are large differences in spatial patterns within and between cancer types.

In this study, we comprehensively evaluated national spatial patterns using the most recent cancer incidence data in China. We focused particularly on how these cancers co-occur and where multiple cancer types are likely to co-occur. We proposed a new research framework to address the challenges of evaluating the high-dimensional spatial co-occurrence of multiple cancer types and treat the contribution of different cancer types equally (regardless of their incidence). We hope that this work will contribute to a more holistic picture of cancer incidence patterns that will provide further targets for tailored prevention and control strategies.

Data source

This study drew on population-based cancer incidence data from 2016 collected by the National Cancer Center in China and published in the 2019 China Cancer Registry Annual Report. This report provides the latest cancer incidence statistics from 487 cancer registry institutes (CRIs), covering 31 provinces (autonomous regions and municipalities) and approximately 381 million inhabitants in mainland China. Details of the number of CRIs and the population coverage in each province are shown in Figure S1(Additional file 1). All new cases diagnosed in 2016 were classified using the International Classification of Diseases 10. [9] The National Cancer Center performed a rigorous data quality control, details of which can be found elsewhere.[10]

We selected 13 cancers with relatively higher incidence/mortality (common cancer) or low incidence but lower 5-year survival (uncommon cancer) in China. These were esophageal, stomach, colorectal, liver, gallbladder, pancreatic, lung, bone, breast, kidney, and brain cancer, leukemia, and lymphoma. [10, 11] Total cancer incidence was then defined as the sum of these 13 cancers.

Research framework for assessing spatial co-occurrence of multiple cancers

The spatial co-occurrence of multiple cancers was assessed using spatial clustering of a single cancer. The global and local Moran’s I for each cancer type was calculated to assess their spatial clustering pattern using the following equations.

$$I=\frac{n}{{\sum\limits_{{i=1}}^{n} {\sum\limits_{{j=1}}^{n} {{w_{ij}}} } }} \times \frac{{\sum\limits_{{i=1}}^{n} {\sum\limits_{{j=1}}^{n} {{w_{ij}}} } ({x_i} - \bar {x})({x_j} - \bar {x})}}{{\sum\limits_{{i=1}}^{n} {{{({x_i} - \bar {x})}^2}} }}$$,

and

$${I_i}=\frac{{{x_i} - \bar {x}}}{{{S_i}^{2}}} \times \sum\limits_{{j=1,j \ne i}}^{n} {{w_{ij}}({x_j} - \bar {x})} ,{\text{ }}{S_i}^{2}=\frac{{\sum\limits_{{j=1,j \ne i}}^{n} {{{({x_j} - \bar {x})}^2}} }}{{n - 1}}$$,

where is the number of CRIs;${x_i}$ the cancer incidence rate for the ${i^{th}}$ CRI ($i=1,2, \cdots ,{\text{ }}n$); ${x_j}$ the cancer incidence rate for ${j^{th}}$ (neighboring) CRI ($j=1,2, \cdots ,{\text{ }}n;i \ne j$);$\bar {x}$the mean incidence of cancer over all CRIs ; and ${w_{ij}}$ the neighboring relations of CRI i and j.

As global Moran’s I moves closer to 1, there is more likely to be spatial clustering of cancer incidence. The p-value shows statistical significance (p < 0.05). Local Moran’s I can identify five spatial clustering patterns for each cancer type. These are high–high (HH) and low–low (LL) clusters (i.e., neighboring CRIs covered area with similar high or low incidences, above or below $\bar {x}$, clustered together); high–low outlier (HL, a CRI covered aera with high incidence surrounded by others with low incidence), low–high outlier (LH, low incidence CRI surrounded by high incidences CRI) and non-significant (NS) clusters. $\bar {x}$ is the mean incidence of the 487 CRIs for cancer over which local Moran’s I is calculated. The five spatial clustering patterns for each cancer were assigned to three hierarchical incidence groups: we defined patterns containing high incidence (HH or HL) to the H group; patterns containing low incidence (LH or LL) to the L group; and NS representing average cancer incidence to the M group.

We further defined three risk areas based on the definition of risk group. For the 13 analyzed cancers, if a CRI contained six or more cancers in the H group, the CRI was considered to be dominated by high-incidence cancer co-occurrence, and its coverage area was defined as a high-risk area. Similarly, if a CRI contained six or more cancers in the L group, the CRI was considered to be dominated by low-incidence cancer co-occurrence, and its coverage area was defined as low-risk. The remaining areas were defined as medium-risk. We assumed co-occurrence of cancers has cumulative effects, which are related to the number of cancers in the same incidence (H/L) group, i.e., where there are more cancers in the H group, the cumulative effect of cancer co-occurrence will be stronger, and vice versa. The overview of this research framework is shown in Figure S2 (Additional file 1).

Statistical analysis

We used a negative binomial regression to obtain the mean incidence over 487 CRIs, together with the incidence rate ratio (RR) and its 95% confidence interval (CI) for each stratified cancer. The clustering of each cancer pair within risk areas was assessed using the Pearson correlation. The population attributable fraction (PAF) and expected excess cases (EC) were used to estimate the proportion of cancer incidence attributable to different risk areas. The PAF was calculated by:

$PA{F_x}=\frac{{{p_x}\left( {R{R_x} - 1} \right)}}{{{p_x}\left( {R{R_x} - 1} \right)+1}}$

${p_x}=\frac{{{\text{population in }}x{\text{ risk area}}}}{{{\text{ population in }}x{\text{ risk area}}+{\text{population in low-risk area}}}}$

where x denotes high- or medium-risk areas; and RR_x the rate ratio of total cancer in x risk areas. The EC attributed to the risk area was calculated by:

${\text{EC}}=\left( {IC{\text{ in }}x{\text{ risk area}}+IC{\text{ in low-risk area}}} \right) \times PAF$

where IC is the number of total cancer incidence cases. ArcGIS software (version 10.6) was used to calculate the global and local Moran’s I, and to visualize all maps. All other analyses used SAS software (version 9.4).

Table 1 shows the mean incidence (per 100,000 population) of 13 cancers in 487 CRIs. This highlights the variation in incidence. The mean incidence varied strongly between cancers, with the top three being lung, liver, and stomach cancers (56.0, 29.1, and 29.0) and the bottom three being kidney, gallbladder, and bone cancers (4.1, 3.6, and 2.0). There were also large spatial differences within the same cancer type, e.g., the highest incidence of lung cancer was 16 times higher than the lowest incidence (151.76 vs. 9.51), and the highest incidence of bone cancer was 28 times higher than the lowest incidence (6.23 vs 0.22) across all CRIs.

Spatial clustering of single cancer type

From a spatial perspective, global Moran’s I showed that there was spatial clustering of incidence for all the cancer types (P < 0.05, Table S1, Additional file 1). Dramatic spatial variation could be observed within each cancer (Fig. 1). Figure 2 shows the mean incidence and the incidence rate ratio (RR) of the five spatial patterns (i.e., HH, HL, NS, LH, LL), which differed significantly within each cancer. For all cancers analyzed, the RR showed a similar dose–response relationship with spatial clustering patterns, i.e., RR decreased from H to M and L groups. The largest variation of RRs was in esophageal cancer, where the highest value was 6.79 times the lowest. The smallest variation was in lung cancer, where the highest was 2.07 times the lowest (Fig. 2). Some correlations between cancer pairs were universal, such as the Pearson coefficients of 0.45 between lung and breast cancers and 0.47 between lung and kidney cancers (Figure S3, Additional file1). There was also some degree of overlap in terms of spatial distribution among the cancer types.

Figure 1 shows the spatial clustering patterns of all the analyzed cancers, showing both large variations and a degree of similarity. Figure 1a shows that the distribution of spatial clustering patterns of lung cancer was very similar to breast cancer. The HH clusters were mainly distributed in coastal and northern provinces, and 65.8% of lung cancer HH clusters overlapped with 62.9% of breast cancer HH clusters. Figure 1b shows the distribution of spatial clustering patterns of digestive system cancers. The most obvious feature is that the HH clusters of esophageal, stomach, and gallbladder cancers were mainly distributed in the east–west belt. There was a strong correlation between esophageal and stomach cancer (Pearson coefficient 0.58, Figure S3, Additional file 1), with 65.4% and 62.6% of HH clusters overlapping each other. Overall, 49.6% and 64.5% of HH clusters overlapped for stomach and gallbladder cancers (Pearson coefficient of 0.21).

Figure 1c shows the spatial clustering patterns of the five relatively uncommon cancers. In addition to the very similar clustering pattern with pancreatic cancer, kidney cancer was also very similar to breast cancer and lung cancer, and the Pearson coefficients ranged from 0.42 to 0.65 between the four cancers. The highest overlap was between breast cancer and kidney cancer, with 80.45% of kidney cancer HH clusters overlapping with 59.3% of breast cancer HH clusters.

For two hematological cancers, lymphoma and leukemia, the HH clusters were mainly located in the southeast (98.3% and 89.5%) and they were strongly correlated (Pearson correlation of 0.60). For brain and bone cancer, NS clusters predominated, indicating a relatively sporadic state, especially for bone cancer (accounting for 66.1% of the CRIs). It was not strongly correlated with any other cancers.

Spatial co-occurrence of multiple cancer types

Figure 3a shows the distribution of high-, medium-, and low-risk areas for the co-occurrence of multiple cancers. Overall, 83 areas (17.1%) were high-risk, 254 (52.2%) medium-risk, and 149 (30.7%) low-risk. High-risk areas were mainly located in the east coastal (Shandong, Shanghai, Jiangsu, and Zhejiang) and northeast (Jilin and Liaoning) areas. Low-risk areas were mainly located in the south, and medium-risk areas were scattered throughout mainland China. The mean total cancer incidence was 282, 208, and 175 per 100,000 population in each type of area, and the RRs (95% CI) of high- and medium-risk areas compared with low-risk areas were 1.61 (1.53–1.69) and 1.19 (1.14–1.23), with no overlap in the 95% CI (Fig. 3a).

The plausibility of the definition was further validated by kvthe observed monotonically increasing or decreasing trend of cancer incidence within high-risk and low-risk areas. The mean incidence of total cancer increased by 22 cases per 100,000 population for each increase in the number of cancers in the H group (i.e. when six, seven, eight, and more than eight of 13 cancers were identified as H group) in high-risk areas regardless of whether the cancer was common or uncommon. In low-risk areas, the mean incidence of total cancer decreased by 11 cases per 100,000 population for each increase in the number of cancers in the L group (P < 0.05, Figure S4, Additional file 1). This suggests that the spatial co-occurrence of cancers has a cumulative effect.

The proportion of the CRIs in different risk-type areas of spatial co-occurrence for multiple cancers in each province is shown in Fig. 3b. Except in the provinces with fewer CRIs (Shanghai, Beijing, Tibet, etc.), the CRIs in most provinces did not occur in only one kind of risk area. Zhejiang had the highest proportion of high-risk areas (85.7%), and Hunan the highest proportion of low-risk areas (86.4%).

Table S2a–S2c(Additional file 1) shows the proportions of CRIs in each of the three hierarchical incidence groups (H, M, and L groups) for each cancer type in each province. As the proportion increases, there is more likelihood that cancers co-occurred in the same CRI. We therefore selected cancers with proportions ≥ 50% to visualize the co-occurrence pattern. Figure 4a–c shows the detailed cancer co-occurrence pattern for each province across different risk areas.

In general, the co-occurrence of cancers was ubiquitous regardless of risk-type areas but the types of cancer involved varied greatly. In the high-risk areas (Fig. 4a), 10 types of cancers co-occurred in Shanghai and Zhejiang, but only one pair of cancers co-occurred in Chongqing and Hubei (lung and lymphoma). In the high-risk areas, lung, breast, colorectal, pancreatic, and kidney cancers were the most frequently observed cancers in the northeast, and lung, breast, stomach, gallbladder, and pancreatic cancers in the eastern coastal region. In the 15 provinces with high-risk areas, lung and pancreatic cancers occurred most frequently together (in nine provinces), followed by breast, stomach, colorectal, gallbladder, and kidney cancers (in seven provinces). For the 30 provinces (except Shanghai) with medium-risk areas, the most frequently observed cancers were brain cancer, bone cancer, leukemia, and gallbladder cancer, which occurred in at least 16 provinces (Fig. 4b). In low-risk areas, the co-occurrence pattern containing lung, colorectal, and liver cancers and lymphoma was the most frequently observed in the 12 northern provinces. The co-occurrence pattern containing esophageal, stomach, gallbladder, pancreatic, and kidney cancer was the most frequently observed in the 12 southern provinces (Fig. 4c).

Table S3 (Additional file 1) shows the PAF of high- and medium-risk areas compared with low-risk areas for each province. The PAF ranged from 7.6–37.9% for high-risk areas and 1.8–15.9% for medium-risk areas. The ECs ranged from 329 (Chongqing) to 32,590 (Jiangsu) for high-risk areas and 18 (Tibet) to 5,442 (Chongqing). The highest total number of ECs was in Jiangsu and the lowest was in Tibet. The total across all CRIs was 44,568.

Table 1. Basic statistics of incidence rate per 100, 000 of analyzed cancers.

Cancer site	ICD-10	N	Mean	Std	Minimum	p25	p50	p75	Maximum
Total		487	210·69	53·53	55·13	170·53	199·23	242·55	450·98
Lung	C33-34	487	55·99	18·86	9·51	43·50	52·09	67·71	151·76
Liver	C22	487	29·10	11·84	5·16	21·90	27·22	32·99	122·02
Stomach	C16	487	28·97	18·92	3·58	15·92	24·55	36·13	131·37
Colorectal	C18-21	487	25·67	10·47	1·90	18·05	23·46	32·06	74·11
Esophageal	C15	486	19·49	17·76	0·60	7·40	13·41	25·58	103·38
Breast	C50	487	17·81	7·97	1·38	12·36	16·26	22·73	46·88
Brain	C70-72, D32-33, D42-43	487	7·09	3·05	1·06	4·99	6·81	8·73	21·37
Pancreatic	C25	487	6·44	4·16	0·34	3·84	5·34	8·13	46·63
Leukemia	C81-86,88,90,96	470	5·54	2·55	0·22	3·90	5·36	7·15	19·93
Lymphoma	C91-95, D45-47	466	5·46	3·01	0·19	3·15	5·17	7·16	19·06
Kidney	C64-66,68	483	4·08	2·78	0·25	2·18	3·37	5·12	16·89
Gallbladder	C23-24	481	3·56	2·18	0·29	1·93	3·08	4·67	12·87
Bone	C40-41	485	2·02	1·03	0·22	1·27	1·86	2·53	6·23

This study drew on the most recent data from cancer registries in China, and proposed a new research framework for investigating the pattern of spatial co-occurrence of 13 cancers. We identified three grades of risk areas and localized them based on the spatial clustering of each cancer type. Our results suggest that areas at high risk of cancer co-occurrence are mainly located in the east and northeast regions and low-risk areas in the south region. We identified the proportions and co-occurrence patterns of the three risk areas in 31 provinces to enable local prioritization. These indicate that the phenomenon of cancer co-occurrence is spatially universal regardless of the level of cancer incidence or the level of risk in the area. These results could provide evidence for priority allocation of resources and developing tailored prevention and control strategies. The framework can be easily adapted to any cancer types, and could therefore serve as a reference for studying cancer co-occurrence in other countries.

Over the past three decades, China has witnessed a transition in its cancer spectrum. In the 1990s, stomach, liver, lung, and esophageal cancers were the top four causes of the disability-adjusted life-years burden, and lung cancer became the top cause in 2019. The incidence of colorectal cancer, which is a greater burden in developed countries (e.g. the United States and the United Kingdom),[12] is increasing in both men and women in China,[13] with the burden rising from sixth to third. Breast and pancreatic cancers, which used to have a relatively low burden (ninth and twelfth in 1990), have now risen to sixth and seventh. Esophageal and stomach cancer remain important (second and fourth in terms of disability-adjusted life-years burden in 2019),[12] It is therefore obvious that China is moving from developing to developed country and facing the coexistence of two cancer spectra. This suggests that regions with different levels of economic development may face challenges from different types of cancer.

In our study, co-occurrence of cancer was observed across the country, but the specific cancers in the pattern of co-occurrence varied between high-, medium-, and low-risk areas. In high-risk areas, the most frequently observed cancers in co-occurrence patterns were lung and pancreatic cancer. This echoes the results of a European study[14] and may be partly explained by the rates of smoking as an important common risk factor,[15] especially among current smokers in the higher age groups, where tobacco causes a more serious DNA damage.[16] Heavy alcohol consumption, low levels of activity, high fasting glucose, and high body mass index can all increase the risk of pancreatic cancer.[15, 17] Focusing on factors that affect lung cancer may therefore also be effective against pancreatic cancer, which has relatively low incidence but high fatality and is positively correlated with economic development. The highest incidence is seen in countries with high socioeconomic status, e.g. the United States, the United Kingdom, and France.[18] It has also been reported that the mortality of pancreatic cancer in Europe is increasing year on year, especially among those over 65 years old.[15] The burden of pancreatic cancer in China therefore cannot be ignored in the context of population aging and economic change, and precautions must be taken in advance. Exploring the common etiology between lung cancer and pancreatic cancer will help the prevention and control of cancer, reduce costs, and improve efficiency.

In medium-risk areas, brain tumors, bone cancer, and leukemia are the most frequently observed cancers. Together with gallbladder cancer, they are the most lethal cancers in children.[19] One study examined the association of these three cancers with parental occupational social contact, which was a proxy for exposure to infectious agents.[20] It found that high parental social contact is positively associated with the risk of brain cancer in children (astrocytoma), but was negatively associated with lymphocytic leukemia and bone cancer. A more general factor was air pollution [21] and body mass index.[22] A Spanish prospective cohort study found that high body mass index increased the risk of leukemia, and gallbladder cancer. [22] Further investigation of the common causes of cancer co-occurrence in medium-risk areas could help to control incidence and prevent it from rising from moderate to high levels, as well as preventing further escalation in high-risk areas.

In the low-risk areas, stomach and esophageal cancers were the two most frequently observed cancers in the co-occurrence pattern in southern provinces. The HH clusters of these two cancers are mainly distributed in the east–west belt of northern China (Fig. 1b), which is broadly consistent with the results of other studies [19, 23] and also lies in the “Asia Esophageal Belt”. In 2005, China first implemented screening for cancers of the digestive system (upper gastrointestinal, liver, and colorectal cancers), lung cancer, and cervical cancer in rural areas with high prevalence.[24] A population-based multicenter cohort study evaluated the effectiveness of the endoscopic screening program, and reported a greater reduction in incidence and mortality of upper gastrointestinal cancers in the screened group than the unscreened group in high-incidence areas during 2005–2015.[25] However, the incidence of stomach and esophageal cancers is still higher in this area than elsewhere. Smoking, alcohol consumption, and diet have been shown to increase the risk of esophageal cancer, and a case–control study in Shanxi showed that the effect of smoking and alcohol were small.[26] This suggests that there are other important risk factors in this region. The study found that esophageal cancer mortality was higher in regions with a higher drought index,[27] which may be partly because of the high intake of corn and consumption of fewer fruits and vegetables, as well as the high nitrosamine content of food that is caused by drought.[28]

The co-occurrence of various cancers has common basic causes. Cancer development shares eight hallmarks,[4] i.e., sustaining proliferative signaling, evading growth suppressors, resisting cell death, facilitation of replicative immortality, inducing angiogenesis, activating invasion and metastasis, reprogramming of energy metabolism, and evading immune destruction. However, there are common risk factors for the co-occurrence of cancers in the same region. there are also common risk factors for the co-occurrence of cancers in the same region. A previous study reported that the Yangtze River Delta and the Pearl River Delta, the two largest city clusters in China, are hotspots for at least seven types of cancer,[29] which is consistent with the high-risk areas of cancer co-occurrence in our study. Shandong and Northeast China are also at high risk of cancer co-occurrence. This may be associated with socioeconomic status. Both the two city clusters and Shandong have a high level of economic development and population density, with GDP and GDP per capita in 2021 among the top 10 in China (except Anhui)[30]. This brings higher living standards and abundant medical resources, but also leads to environmental pollution and is associated with an older population. [31, 32] These factors are also important in cancer, especially air pollution. Fine particulate pollution (PM2.5) is a risk factor for various cancers, including acute leukemia and lung, colorectal, stomach, and kidney cancers.[21] Water pollution is positively correlated with mortality from digestive system cancers.[33] For decades, the International Agency for Research on Cancer (IARC) has published a list of more than 1000 potential carcinogens, but only 122 are definitive, [34], and progress on identifying carcinogens has been slow. Exploring the spatial co-occurrence of multiple cancers could provide a way of identifying common risk factors. Previous studies have generally investigated the spatial co-occurrence of diseases by using bivariate local Moran’s I to investigate the spatial correlation between two cancers, [35] searching for overlapping areas of high incidence/prevalence of two or three diseases, [29, 36] and traditional cluster analysis [37, 38]. Zhang et al. applied Gi* analysis to identify hotspots of 11 cancers and show how they overlapped in China. However, they did not report cancer co-occurrence patterns in detail.[29]

The framework we propose has some advantages. It provides an explicit way of translating the spatial overlap of multiple cancers with dramatic incidence variation into a risk of co-occurrence. Instead of focusing only on high-incidence cancers in the cancer spectrum, this framework can be used for any cancer type, and the characteristics uncovered could provide insights into the association between any groups or cancers to accelerate knowledge and research on the mechanisms of co-occurrence. The characteristics of areas with different risk levels for the co-occurrence of multiple cancers can be accurately captured, providing a target for prevention and control of multiple cancers by local governments. The distribution of co-occurrence of cancers in our study is consistent with the cancer distribution found in previous studies, supporting the validity of our research framework. Since 1942, when the world’s first population-based cancer registry was established in Denmark, registries have spread to almost all countries. They provide essential data for decisions about actions to control cancer and which groups to target.[39] The consistency of the data platforms used makes it easy to apply our research framework to other countries around the world to explore cancer co-occurrence and to use cancer registry data for further information. The framework could also be applied to other health data to study co-occurrence.

This study also had some limitations. The spatial co-occurrence of cancers was based on CRI incidence data, which could be affected by aspects such as differences in diagnostic techniques among hospitals and cancer registration data quality. Second, there are still insufficient data from areas of western China, such as Qinghai, Xinjiang, and Tibet, meaning the characteristics of cancer co-occurrence in these provinces are not fully reflected.

In summary, the co-occurrence of different cancers was universal. Using the design framework, we can efficiently identify and locate the risk areas with different degree of cancer co-occurrence and specific co-occurrence pattern in each province, which can improve the efficiency of identifying more unknown common risk factors and guide practice.

CRIs, cancer registry institutes; HH, high-high cluster; HL, high-low outlier; LH, low-high outlier; LL, low-low cluster; NS, non-significant; RR, rate ratio; CI, confidence interval; PAF, population attributable fraction; EC, excess cases; IC, the number of total cancer incidence cases; IARC, International Agency for Research on Cancer.

Ethics approval and consent to participate

This study did not conduct data collection, and just analyzed the existed incidence data extracted from 2019 China Cancer Registry Annual Report published by the National Central Cancer Registry of China. The incident cases’ name and ID could not be identified in the dataset. No administrative permissions were required to access and use the dataset in this study. So it is not applicable for ethics approval and consent to participate.

Consent for publication

Not applicable.

Competing interests

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Funding

This study was supported by the CAMS Innovation Fund for Medical Sciences (grant number 2017-I2M-1-009), and the CAMS Innovation Fund for Medical Sciences (grant no: 2021-1-I2M-022).

Authors’ Contributions

JJ designed the study and critically revised the manuscript. WH and CY prepared the material and performed the data analysis. PW , FX and WC verified the data and analytical method. WH and CY prepared the manuscript. WG, YS, LZ, XG and JZ proofed the manuscript. All authors critically revised the manuscript for important intellectual content, contributed to the article, and approved the submitted version.

Acknowledgments

We appreciate the work by the National Cancer Center (NCC).

Availability of data and materials

The data that support the findings of this study are openly available in the published 2019 China Cancer Registry Annual Report by the National Cancer Center (ISBN: 9787117309707). This is a print edition that can be accessed from the National Library of China at the following link: http://find.nlc.cn/search/showDocDetails?docId=-7883337881190403060&dataSource=ucs01&query=2019%E4%B8%AD%E5%9B%BD%E8%82%BF%E7%98%A4%E7%99%BB%E8%AE%B0%E5%B9%B4%E6%8A%A5

Summary tables of. mortality estimates by cause, age and sex, globally and by region, 2000–2019 [https://www.who.int/data/gho/data/themes/mortality-and-global-health-estimates/ghe-leading-causes-of-death].
Sung H, Ferlay J, Siegel RL, Laversanne M, Soerjomataram I, Jemal A, Bray F. Global Cancer Statistics 2020: GLOBOCAN Estimates of Incidence and Mortality Worldwide for 36 Cancers in 185 Countries. Cancer J Clin. 2021;71(3):209–49.
Soerjomataram I, Bray F. Planning for tomorrow: global cancer incidence and the role of prevention 2020–2070. Nat reviews Clin Oncol. 2021;18(10):663–72.
Hanahan D, Weinberg RA. Hallmarks of cancer: the next generation. Cell. 2011;144(5):646–74.
Islami F, Goding Sauer A, Miller KD, Siegel RL, Fedewa SA, Jacobs EJ, McCullough ML, Patel AV, Ma J, Soerjomataram I, et al. Proportion and number of cancer cases and deaths attributable to potentially modifiable risk factors in the United States. Cancer J Clin. 2018;68(1):31–54.
Chen W, Xia C, Zheng R, Zhou M, Lin C, Zeng H, Zhang S, Wang L, Yang Z, Sun K, et al. Disparities by province, age, and sex in site-specific cancer burden attributable to 23 potentially modifiable risk factors in China: a comparative risk assessment. The Lancet Global health. 2019;7(2):e257–69.
van Eggermond AM, Schaapveld M, Lugtenburg PJ, Krol AD, de Boer JP, Zijlstra JM, Raemaekers JM, Kremer LC, Roesink JM, Louwman MW, et al. Risk of multiple primary malignancies following treatment of Hodgkin lymphoma. Blood. 2014;124(3):319–27. quiz 466.
GLOBOCAN. 2020 Map production [https://gco.iarc.fr/].
International Classification of Diseases., 10th Revision (ICD-10) [https://www.who.int/standards/classifications/classification-of-diseases].
Zheng. R, Zhang. S, Wang. ZH, Sun. K S, Chen. R, Li. L. Wei. W, He. J: Cancer incidence and mortality in China, 2016. J Natl Cancer Cent. 2022;2(1):1–9.
Zeng H, Zheng R, Guo Y, Zhang S, Zou X, Wang N, Zhang L, Tang J, Chen J, Wei K, et al. Cancer survival in China, 2003–2005: a population-based study. Int J Cancer. 2015;136(8):1921–30.
Qiu H, Cao S, Xu R. Cancer incidence, mortality, and burden in China: a time-trend analysis and comparison with the United States and United Kingdom based on the global epidemiological data released in 2020. Cancer Commun (London England). 2021;41(10):1037–48.
Xia C, Dong X, Li H, Cao M, Sun D, He S, Yang F, Yan X, Zhang S, Li N, et al. Cancer statistics in China and United States, 2022: profiles, trends, and determinants. Chin Med J. 2022;135(5):584–90.
Boffetta P, Castaing M, Brennan P. A geographic correlation study of the incidence of pancreatic and other cancers in Whites. Eur J Epidemiol. 2006;21(1):39–46.
Carioli G, Malvezzi M, Bertuccio P, Boffetta P, Levi F, La Vecchia C, Negri E. European cancer mortality predictions for the year 2021 with focus on pancreatic and female lung cancer. Annals of oncology: official journal of the European Society for Medical Oncology. 2021;32(4):478–87.
Yoshida K, Gowers KHC, Lee-Six H, Chandrasekharan DP, Coorens T, Maughan EF, Beal K, Menzies A, Millar FR, Anderson E, et al. Tobacco smoking and somatic mutations in human bronchial epithelium. Nature. 2020;578(7794):266–72.
GBD 2017 Pancreatic Cancer Collaborators. The global, regional, and national burden of pancreatic cancer and its attributable risk factors in 195 countries and territories, 1990–2017: a systematic analysis for the Global Burden of Disease Study 2017. lancet Gastroenterol Hepatol. 2019;4(12):934–47.
Lin L, Li Z, Yan L, Liu Y, Yang H, Li H. Global, regional, and national cancer incidence and death for 29 cancer groups in 2019 and trends analysis of the global cancer burden, 1990–2019. J Hematol Oncol. 2021;14(1):197.
Zheng R, Zhang S, Zeng H, Wang S, Sun K, Chen R, Li, L, Wei. W, He; J: Cancer incidence and mortality in China, 2016. J Natl Cancer Cent. 2022;2(1):1–9.
Omidakhsh N, Hansen J, Ritz B, Olsen J, Heck JE. High parental occupational social contact and risk of childhood hematopoietic, brain and bone cancers. Cancer Epidemiol. 2019;62:101575.
Turner MC, Andersen ZJ, Baccarelli A, Diver WR, Gapstur SM, Pope CA 3rd, Prada D, Samet J, Thurston G, Cohen A. Outdoor air pollution and cancer: An overview of the current evidence and public health recommendations. CA: a cancer journal for clinicians 2020.
Recalde M, Davila-Batista V, Díaz Y, Leitzmann M, Romieu I, Freisling H, Duarte-Salles T. Body mass index and waist circumference in relation to the risk of 26 types of cancer: a prospective cohort study of 3.5 million adults in Spain. BMC Med. 2021;19(1):10.
Li P, Jing J, Li R, Ge M, Jia P, Hu W, Qi X, Wei WQ, Zhuang G. Upper Gastrointestinal Cancer in China: Spatial Epidemiologic Evidence from Screening Areas. Cancer Prev Res (Philadelphia Pa). 2020;13(11):935–46.
Cao M, Li H, Sun D, He S, Yu Y, Li J, Chen H, Shi J, Ren J, Li N, et al. Cancer screening in China: The current status, challenges, and suggestions. Cancer Lett. 2021;506:120–7.
Chen R, Liu Y, Song G, Li B, Zhao D, Hua Z, Wang X, Li J, Hao C, Zhang L, et al. Effectiveness of one-time endoscopic screening programme in prevention of upper gastrointestinal cancer in China: a multicentre population-based cohort study. Gut. 2021;70(2):251–60.
Gao Y, Hu N, Han XY, Ding T, Giffen C, Goldstein AM, Taylor PR. Risk factors for esophageal and gastric cancers in Shanxi Province, China: a case-control study. Cancer Epidemiol. 2011;35(6):e91–99.
Wu K, Li K. Association between esophageal cancer and drought in China by using Geographic Information System. Environ Int. 2007;33(5):603–8.
Lin K, Shen W, Shen Z, Wu Y, Lu S. Dietary exposure and urinary excretion of total N-nitroso compounds, nitrosamino acids and volatile nitrosamine in inhabitants of high- and low-risk areas for esophageal cancer in southern China. Int J Cancer. 2002;102(3):207–11.
Zhang L, Wan X, Shi R, Gong P, Si Y. Comparing spatial patterns of 11 common cancers in Mainland China. BMC Public Health. 2022;22(1):1551.
China Statisctical Yearbook. [http://www.stats.gov.cn/tjsj/ndsj/].
Wang Y, Gao W, Song SW, Gong T, Z DJ, Wang L, Liu Z, Tang G, Huo, Y. Contrasting trends of PM2.5 and surface-ozone concentrations in China from 2013 to 2017. Nationall Sci Rev. 2020;7(8):1331–9.
Wang H, Li J, Gao M, Chan TC, Gao Z, Zhang M, Li Y, Gu Y, Chen A, Yang Y, et al. Spatiotemporal variability in long-term population exposure to PM(2.5) and lung cancer mortality attributable to PM(2.5) across the Yangtze River Delta (YRD) region over 2010–2016: A multistage approach. Chemosphere. 2020;257:127153.
Ren H, Wan X, Yang F, Shi X, Xu J, Zhuang D, Yang G. Association between changing mortality of digestive tract cancers and water pollution: a case study in the Huai River Basin, China. Int J Environ Res Public Health. 2014;12(1):214–26.
Agents Classified by the IARC Monographs., Volumes 1–132 [https://monographs.iarc.who.int/agents-classified-by-the-iarc/].
Aturinde A, Farnaghi M, Pilesjö P, Mansourian A. Spatial analysis of HIV-TB co-clustering in Uganda. BMC Infect Dis. 2019;19(1):612.
Gwitira I, Murwira A, Mberikunashe J, Masocha M. Spatial overlaps in the distribution of HIV/AIDS and malaria in Zimbabwe. BMC Infect Dis. 2018;18(1):598.
Bhakkan-Mambir B, Deloumeaux J, Luce D. Geographical variations of cancer incidence in Guadeloupe, French West Indies. BMC Cancer. 2022;22(1):783.
Cassetti T, La Rosa F, Rossi L, D'Alò D, Stracci F. Cancer incidence in men: a cluster analysis of spatial patterns. BMC Cancer. 2008;8:344.
Wagner G. History of cancer registration. IARC Sci Publ 1991(95):3–6.

No competing interests reported.

Supplementarytableandfigures.pdf

Download PDF

Version 1

posted

You are reading this latest preprint version

Spatial co-occurrence of multiple cancers: a population-based study using national cancer registry data in China

Status:

Version 1

Abstract

Background

Methods

Results

Conclusion

Figures

Introduction

Materials and methods

Statistical analysis

Results

Discussion

Conclusion

Abbreviations

Declarations

References

Additional Declarations

Supplementary Files

Status:

Version 1