Studies of the geographical distribution of cancer mortality in Spain have revealed the existence of different spatial patterns for different cancer sites that is difficult to explain. The aetiology of malignant tumours is of great complexity, owing to the presence of many determinants of a different nature (environmental, including habits, diet, environs and occupation, biological and genetic), some of which (environmental) are and some of which (biological and genetic) are not linked to the territory.
The results of this study suggest that low bioavailable arsenic levels in soil might give rise to a population exposure that was statistically associated with higher mortality due to cancers of the stomach, pancreas, lung and brain and NHL, among men and women alike. While chromium topsoil levels were associated with higher female mortality due to cancers of the upper gastrointestinal tract (buccal cavity, pharynx and oesophagus), breast cancer and NHL, no such association was found in men.
Arsenic is a known carcinogen in the skin, lung, bladder, liver and kidney, with the evidence suggesting that lung cancer is the most common cause of arsenic-related mortality (IARC 2012). People can be exposed to arsenic in food and water and from inhalation, e.g. breathing sawdust or smoke from burning arsenic-treated wood or fly ash from combustion of As-rich coal (ATSDR 2007).
Current evidence indicating that exposure to arsenic is a risk factor for cancer in the general population comes from occupational studies based on cohorts of workers who inhaled air contaminated by arsenic and other products and from studies in places with populations exposed to high arsenic concentrations in drinking water over prolonged periods of time (Straif et al. 2009). These have highlighted its association with the increase in incidence of lung, bladder, skin, kidney, liver and possibly prostate cancer (Nordstrom 2002). Currently, the greatest interest in the toxicology of arsenic lies in exposure deriving from this substance’s natural presence in food, water and soil. Understanding the environmental levels that could cause public health problems is thus a critical research area (Hughes et al. 2011).
In the USA, a nationwide survey conducted in areas that were judged not to have anthropogenic sources of arsenic reported that natural background concentrations in soil ranged from less than 1 to 97 mg kg−1 (Shacklette and Boerngen 1984). According to our study data, the range was very similar, i.e. 1 to 99.4 mg kg−1 (Locutura et al. 2012). Owing to low arsenic bioavailability in soil, it is believed that, as compared to intake of naturally occurring arsenic in water and diet, soil arsenic constitutes only a small fraction of intake (Boyce et al. 2008). In the US population, the major food contributors to inorganic As exposure were the following: vegetables (24 %); fruit juices and fruits (18 %); rice (17 %); beer and wine (12 %); and flour, corn and wheat (11 %). Approximately 10 % of total As exposure from foods is in the form of toxic inorganic As (Xue et al. 2010).
Furthermore, the concentration of heavy metals in soil also determines their presence in animal tissue (López Alonso et al. 2002), and the use of biomarkers in cattle has been suggested as a way of monitoring these elements in the environment, since they avoid the problem of bioavailability posed by soil samples.
The small number of studies means that there is very little epidemiological evidence of the association between arsenic topsoil levels and frequency of cancer. However, heavy metal and arsenic topsoil concentrations serve as an indicator of long-term exposure to these elements (Tchounwou et al. 2012). A recent study on arsenic topsoil levels and cancer undertaken in a province in China reported an association with mortality due to cancers of the colon, stomach, kidney, lung and nasopharynx (Chen et al. 2015): this study included 83 towns, 1683 top soil samples and mortality across the period 2005–2010. Although the dimensions of our study were very different, in view of the fact that it covered a 10-year mortality period from 1999 to 2008, included all 7917 towns across mainland Spain and used 13,317 sampling points in estimating arsenic and chromium levels, there is a certain coincidence in terms of the tumour sites for which excess risk was found.
Numerous studies have identified associations between lung cancer and inhaled hexavalent chromium (Cr(VI)) in occupational settings. Furthermore, it is a component of the carcinogenicity of tobacco smoke. Chromium may possibly cause gastrointestinal tract cancer due to drinking Cr-laden water and eating Cr-laden vegetables (Peralta-Videa et al. 2009; Welling et al. 2015). Inhalation of Cr(VI) has occurred in a number of industries, including leather tanning, chrome plating, cement works and stainless steel welding and manufacturing.
It is noteworthy that, in addition to breast cancer, our study observed the association between chromium concentrations and cancers of the buccal cavity, pharynx and oesophagus and NHL exclusively in women. To our knowledge, the origin of this differential risk is unknown, though in the case of cancer of oesophagus, it might be linked to the different geographical mortality pattern. One possible explanation could be that exposure to food and drinking water containing chromium has greater toxicity because it can take place over the long term (e.g., lifetime) and is more likely to occur at particularly susceptible life stages (e.g., in foetuses, children and pregnant women) than in occupational exposures (Welling et al. 2015).
Heavy metal pollution in soil has received much attention because metals are hardly decomposable by soil microbes and can amplify with food chain extension, which in turn poses a potential threat to human health (Li et al. 2014). Human beings could be exposed to heavy metals from vegetable soils via the following six main pathways: (1) direct ingestion of soil particles, (2) dermal contact with soil particles, (3) diet through the food chain, (4) inhalation of soil particles from the air, (5) oral intake from groundwater and (6) dermal intake from groundwater (Abrahams 2002; Liu et al. 2013).
We are unaware of the existence of any study comparable to ours in terms of dimension and scope. Our study encompasses the whole of mainland Spain, contains an estimate of As and Cr topsoil levels for close on 8000 towns obtained from a mesh of more than 13,000 sampling points and covered a broad study period spanning mortality over 10 years. Statistical analysis was performed using hierarchical models with a spatial component (Besag et al. 1991) fitted by R-INLA (Lindgren and Rue 2015). In these models, the risk of falling into the ecological fallacy is minimised by using a very small spatial scale and making no inferences at an individual level (Clayton et al. 1993). Moreover, to account for the spatial interpolation error in the inference, a multivariate model for spatially misaligned data is used (the set of observed locations for the explanatory variable is not identical to that for the response variable) (Cameletti et al. 2013). In this model, the inference is arrived at using the SPDE approach (Lindgren et al. 2011), which makes it computationally feasible and efficient. Although this model only allows to estimate the RR of the variable of exposure as a continuous variable, the estimation in many cases confirms the results of previous analyses and, being more conservative, generally going in the same direction of the association.
Data from soil geochemical studies are usually recorded in parts per million or milligram per kilogram and have been called compositional data requiring specific transformations (Aitchison 1994, 2003). We recognise the compositional/multivariate inherent soil data nature, but this aspect has not been explored in this study. This line has to be developed to a greater extent.
Insofar as limitations are concerned, it should be noted that this was an ecological mortality study with all the problems of using data grouped by town. The study assumed that As and Cr topsoil levels determined each town’s population exposure, and data on possible important confounding variables, such as smoking habit, were lacking. Even so, an effort was made to control for such confounders, by including a series of socio-demographic components as variables of adjustment and by attempting to control for the anthropogenic origin of As and Cr through data on the proximity of the sources of these elements.
Furthermore, it is important to stress that residing in a town with Cr and As levels in the upper quartile in no way implies that their spatial location would in itself give rise to any given cancer. The influence on the population of other socio-demographic and lifestyle factors and other exposures must be borne in mind when it comes to assessing the associations found.
With respect to possible intervention measures, a constant factor when reviewing publications relating to metal and metaloid soil concentration is the warning sounded by researchers as to the importance of controlling and limiting As levels in both soil and, due to its incorporation into the trophic chain, food (Micó et al. 2007; Burló et al. 2012; Muñoz et al. 2000; Peralta-Videa et al. 2009; van Geen et al. 1997; Delgado-Andrade et al. 2003; Peña-Fernández et al. 2014).
To conclude, the results show a statistical association in men and women alike between arsenic topsoil concentration and mortality due to cancers of the stomach, pancreas, lung and brain and NHL. Furthermore, an association was observed with cancers of the buccal cavity and pharynx, colorectal, renal and prostate in men. Chromium topsoil levels were associated with higher mortality in women due to cancer of the upper gastrointestinal tract, breast cancer and NHL, but no such association was found in men.
Access to the data of composition of the soil and its inclusion in epidemiological studies of health in humans is very innovative and opens an important way to try to understand the set of expositions that determine the frequency of cancer and other chronic diseases. On the other hand, the contribution of the geochemical atlas with an entire country geo-coded data is a great contribution to the environmental epidemiology and public health in general.