(324b) Multiscale Connectivity for Chemical Mixture Toxicity Assessment

Toxicity of chemical mixtures is only partially addressed by the current state of environmental health sciences. The so-called 'cocktail' effect becomes all the more complex due to the large number of possible combinations of chemicals and other (physical or biological) stressors in the environment. This has hampered the development of rigorous methodologies for tackling the issue of environmental mixture safety. Currently, our scientific understanding and policy for environmental mixtures are based largely on extrapolating from, and combining, data in the observable range of single chemical toxicity to lower environmental concentrations and composition - i.e. using higher dose data to extrapolate and predict lower dose toxicity. More precise approaches to characterize toxicity of mixtures are needed. A major obstacle to the development of effective mixture risk assessment methodologies is the possibly infinite ways of combining chemicals into actual environmental mixtures. It should be noted, however, that although in theory the number of combinations of chemicals or stressors is infinite, the number of biological processes is finite. Therefore, in considering an integrated approach for risk assessment, it makes more sense to work on the finite biological processes that may be affected by human exposure to these mixtures rather than the infinite combinations of chemicals and stressors.

Integrated health impact assessment of environmental stressor mixtures would need to follow a ‘full chain’ approach to take into account all relevant health stressors and their interaction. Application of the full chain approach entails considering all possible exposure pathways via the environment and lifestyle choices. It also encompasses considering the effects of co-exposure to relevant stressors and how risk modifying factors such as age, diet, gender, and time window of exposure affects the final physiological response. It is clear that successful application of this approach poses demanding data requirements both in terms of environmental monitoring and in terms of biological and clinical data interpretation. What is most important is the need for comprehensive data interpretation of the molecular, biochemical and physiological processes that couple exposure to health outcome.

This requires forging a new paradigm for interdisciplinary scientific work in the area of environment and health. We shall call this the connectivity paradigm for chemical risk assessment, denoting an approach that builds on the exploration of the interconnections between the co-existence of multiple stressors and the different scales of biological organisation that together produce the final adverse health effect. Connectivity marks a clear departure from the conventional paradigm, which seeks to shed light on the identification of singular cause-effect relationships between stressors and health outcomes. It entails creating a new way of combining health-relevant information coming from different disciplines, including (but not limited to) environmental science, epidemiology, toxicology, physiology, molecular biology, biochemistry, mathematics and computer science (Kitano, 2002). In this paper we discuss the integration of these different information classes into a unique framework to better inform and support public health impact assessment of chemical mixtures in the environment.

In this context, we move forward towards the definition of a co-exposure biology using optimally physiology and systems biology based modeling and data mining and assimilation algorithms. The aim is to couple the “systems biology” approach to the gene-environment interactions with the corresponding “physiome” approach. The variable levels of biological organization involved in this holistic view of mixture toxicology and cumulative exposure and risk assessment suggest that different technologies need to be brought to bear in order to obtain a comprehensive view of how co-exposure to multiple chemicals affects the overall phenotypic response of individuals. Technological variability introduces the need for better data integration and assimilation and for the development of novel data analysis and hypothesis generation and testing procedures in order to best elucidate the biological mechanisms underlying mixture toxicity. On the basis of the data currently available, the best available techniques, and the need for data integration, a tiered approach has been developed and outlined herein to support the connectivity approach to the assessment of health risk associated to combined exposure to multiple chemical stressors. This exposure biology approach to mechanism-based risk assessment of environmental chemical mixtures can be tackled with an integrated, multi-layer computational methodology, ideally comprising the following steps:

a) Characterization of exposure factors quantifying the parameters that affect human exposure to environmental chemicals, such as time-activity relationships, seasonal and climatic variation, and consumer choice. These exposure factors can be used to derive aggregate and cumulative exposure models, leading in probabilistic exposure assessments. Aggregation can be done across exposure pathways and routes and even across different exposure scenarios, if the relevant exposure metric or the imputable biological or physiological effect can be related to these scenarios. For instance, exposure to volatile organic compounds (VOCs) such as benzene or toluene and mixtures thereof may occur both from environmental media and in specific occupational settings. A cumulative exposure scenario for these substances would have to take stock of the actual variability of exposure across these different settings throughout typical days for the same period in an individual’s lifespan.

b) Current toxicological state of the art combines estimations of biologically effective dose with early biological events to derive dose-effect models, which can be used in combination with the probabilistic exposure estimates to derive biomarkers of exposure and/or effect. Combined use of epidemiological, clinical and genetic analysis data may shed light on the effect of risk modifying factors such as lifestyle choices and DNA polymorphisms. Observation of real clinical data and /or results of biomonitoring, if coupled with the exposure/effect biomarker discovery systems, can produce biomarkers of individual susceptibility and thus allow estimations of individual response to toxic insults. Toxicogenomics, comprising transcriptomics, proteomics and metabolomics, and adductomics (considering adducts of xenobiotics not only to DNA but also to proteins such as albumin) are key technologies to this kind of analytical and data interpretation process. 

c) The analysis of the biomarker data (including results on biomarkers of exposure, effects and individual susceptibility) results in the integrated assessment of risk factors. Use of information on risk factors with molecular dosimetry data (i.e. estimation of the actual internal and biologically effective dose of xenobiotic substance found in the target organ and, indeed, perturbing cellular response) enables population risk studies to be done, by converting generic exposure profiles into population risk metrics having taken into account inter-individual variability of response and exposure uncertainty. 

The connectivity approach was applied on data from a Europe-wide campaign on environmental and biological monitoring of a virtually ubiquitous mixture of volatile organic compounds, i.e. benzene, ethylbenzene, toluene and o-, m- and p-xylene, aldehydes such as formaldehyde and acetaldehyde and a complex mixture of polyaromatic hydrocarbons, which is typical of combustion products. The full array of –omics technologies outlined above were applied to samples of indoor air and dry blood spots and urine of exposed subjects from almost all European Union capitals. Exposure assessment was completed with detailed time activity diaries and questionnaires regarding smoking and dietary habits (especially with regard to alcohol consumption). Gene expression results were processed using clustering algorithms to derive heat maps demonstrating clearly the differences in the biological perturbations caused by parts of the indoor and ambient airborne chemical mixtures in the sampled sites.  Pathway analysis using the PANTHER system on-line showed that two key pathways, the p53 and oxidative stress were differentially modulated from specific chemical families such as aldehydes, while specific genes or gene sequences could be characterized as molecular markers of exposure. Following through to the biological processes that were perturbed by exposure to these airborne chemical mixtures we found that during acute (short-term) exposure signal transduction and mRNA transcription were modulated the most following an inverse dose-response function. When chronic (longer-term) exposure results were analyzed, protein metabolism, mRNA transcription regulation and cell proliferation and differentiation were the main mechanisms that were modulated following a normal dose-response behavior. In addition, the mixtures that were richer with non-carcinogens were the ones that had the highest impact on the biological processes induced during short-term exposure; on the contrary, mixtures richer with carcinogens such as benzene induced a higher response in terms of biological process induction to the individuals who were exposed longer. This behavior was confirmed by metabolomics profiling, which showed a relative increase in benzene metabolites such as s-mercapturic acid as well as free, non-metabolized benzene. Phenotypic observations confirm that chronic exposure to carcinogenic VOCs such as benzene could increase the risk of leukemia. Taking into account the metabolic processes and interactions (e.g. competitive inhibition) that regulate the effective metabolism of the VOCs to which the European population was exposed we estimated the actual risk of cancer from the combined exposure to such airborne chemical mixtures released from fuel and consumer products.

Following the integrated multiscale connectivity methodology coined herein it is possible to identify the molecular pathways and biological processes that underlie the onset or exacerbation of disease phenotypes associated to co-exposure to chemical mixtures. This approach sheds light to the respective pathways of toxicity and assigns causal associations between environmental stressors and human health.