The exposome in respiratory diseases: multiple preventable risk factors from early life to adulthood

Context

The end of the 20th century and the beginning of the 21st century have been marked by increased global industrialisation and urbanisation, causing profound changes in our lifestyle choices and the living environment. The increase in the prevalence of respiratory diseases over the last few decades has mirrored the urbanisation process, which raises questions about the impact of environmental factors on the lungs [1]. The lungs develop from the embryonic phase until young adulthood [2], making them particularly vulnerable to adverse environmental exposures during this critical developmental period. A growing body of research has advanced our scientific understanding of how exposures occurring during prenatal and childhood periods may have lifelong implications on respiratory health [3]. However, knowledge is still inadequate to fully understand the most critical exposure windows for different environmental exposures such as phthalates, air pollution or green spaces [4, 5]. The classical approach in epidemiology is to study each exposure–outcome relationship separately. While this approach is valuable to generate important evidence for environmental policy-making (e.g. setting threshold-based air quality guidelines), it does not always consider that different exposures are sometimes linked, and complex, from a life-course perspective. In this context, the “exposome” concept has been a major development in recent years. It was proposed in 2005 by Wild [6], and refers to the wide environmental (i.e. non-genetic) determinants on health and diseases from conception and onwards. To investigate multiple external exposures–outcome relationships, several types of statistical approaches have been reported in the literature: single-exposure regression-based methods (e.g. exposome-wide association study (ExWAS)), multiple-exposures regression-based methods (e.g. weighted quantile sum (WQS) regression), supervised clustering approaches (e.g. latent class analysis) or analysis accounting for the hierarchical structure of the data (e.g. Bayesian kernel machine regression) [7]. These methods can be combined, for example, by applying an ExWAS followed by a cluster analysis on a restricted set of the relevant exposures from ExWAS results [7]. These approaches are complemented by analyses of a wide range of biological responses in the internal environment using the “omics” technologies. The exposome research into respiratory diseases is still in its infancy. However, as a wide array of environmental exposure as well as personal biological data are being collected, along with the technological advances, we start to move towards better understanding of the exposome on some major respiratory diseases such as asthma [7].

In this journal club, we put into perspective three publications investigating the effects of environmental exposures considered separately or in an exposome approach, with different exposure windows, on respiratory outcomes. These three studies involved different populations, respiratory outcomes, exposures and statistical methods, which provides a comprehensive view of what can currently be done in environmental epidemiology. Far from showing that there is a preferable design and method, each of these studies presents their strengths and limitations and brings complementary results to better understand the effects of environmental exposures on respiratory health.

Main results

Commentary

These three studies are each innovative and provide complementary knowledge in environmental epidemiology relating to respiratory diseases, covering a wide range of environmental exposures from early life to adulthood. Together, they show that for a better understanding of the aetiology of chronic respiratory diseases, we will need to consider the interplay between multiple factors (e.g. up to 87 potential determinants considered in the NutriNet-Santé study) and different exposure windows (adult versus childhood exposures as in the RHINESSA study, and up to the prenatal period as in the INMA study).

One of the main strengths of the three studies included in this journal club is the extensive characterisation of respiratory health and exposures. In the INMA study, repeated spirometry measures were available, covering four time-points in childhood. This allowed the authors to test whether specific age periods are more sensitive to the effects of gestational exposure to phthalates and to evaluate whether effects persist or remit over time. While the study results are in agreement with several previous studies investigating at a single time-point, which have showed statistically nonsignificant inverse associations or null associations, this study is one of the first to assess the association over multiple time-points throughout childhood (i.e. pre-adolescence). Since then, Foong et al. [11] investigated whether prenatal exposure to phthalates was associated with lung function outcomes at three time-points from early childhood into young adulthood (age 5–22 years) and contrarily found it to be associated with improved lung function. However, the differences in age groups and differing methods of phthalate measurement make direct comparisons difficult.

The RHINESSA study is one of the first studies with such detailed calculations of exposures during different susceptibility windows throughout the entire lifespan of adults, and extensive respiratory health characterisation. This is also one of the first studies to investigate the effects of greenness on respiratory health in adults. The authors hypothesised that the deleterious effect of greenness could be due to pollens. However, as the NDVI does not differentiate the types of vegetations, considering pollens and/or using land cover could provide additional information to study the health impacts.

The NutriNet-Santé study is the first study investigating the association between exposome and asthma in a large population of adults. The main strength of the study is the holistic approach, considering simultaneously multiple exposures and early-life exposures, which are critical windows of vulnerability, to explore a multifactorial chronic disease. They identified profiles of combined exposures from early life and lifestyle that were at risk for asthma in adults. One of the difficulties of the exposome approach is the interpretability of the results, which is why it is always interesting to couple their interpretation with studies that focus on a specific exposure. The results are in line with those previously described in the literature; however, a validation cohort would have been interesting to evaluate the validity of the results.

The three studies employ different and complementary statistical methods. The multivariate logistic regressions used in the RHINESSA study are certainly the most classical methods and allow for a simple interpretation of the results, in particular the dose–response relationship. Results based on these classical methods can be more easily compared among different studies. However, it is still difficult to disentangle the individual health impacts within a multi-exposure framework or pollutants mixture using these classical methods. In the INMA study, the use of WQS regression allowed the authors to evaluate the phthalate metabolites mixture effect, leading to an interesting finding that perhaps MBzP could be a main driver of the observed association with reduced lung function during childhood. Novel statistical approaches such as WQS regression could be extremely useful to quantify individual health risk of environmental pollutants mixtures, which will provide targeted public health interventions [12]. Finally, in the NutriNet-Santé study, the authors performed an exposome approach by combining an ExWAS approach with a latent class analysis.

The three studies acknowledged a number of methodological limitations. The INMA study was limited by a relatively small sample size at each time-point, the difficulties in obtaining spirometry measurements in young children and the fact that the included sample differed in some sociodemographic characteristics to those excluded. Another limitation is that childhood exposure to phthalates has not been accounted for.

The response rate of RHINESSA was low (∼40%), which could possibly have led to a selection bias. The authors acknowledged that it was difficult to know which was the most important window of exposure, given that they were highly correlated. Interestingly, for asthma attack in the last 12 months and rhinitis, the most recent exposure windows are those with the strongest associations. This could be explained by fewer measurement errors but also by the fact that these diseases are perhaps more associated with more recent exposures. Although the statistical method used allows a simple interpretation of the results, it may be of interest in future analyses in RHINESSA to apply methods similar to those used in the INMA and NutriNet-Santé studies, to address the health impacts from highly correlated environmental variables. The authors did not perform multi-pollutant models between NO₂, PM_2.5, PM₁₀ and black carbon as these pollutants were highly correlated. Multi-pollutant models are complex to perform and interpret given the relationship of air pollutants to each other [13]. However, an analysis considering all pollutants, such as the WQS regression approach performed in the INMA study, could be interesting. In this study, the authors mutually adjusted air pollution and NDVI. As the relationship between these exposures is complex [14], it would be interesting to test their interaction and/or to perform mediation analyses, although this would usually require a relatively large sample size.

The French NutriNet-Santé cohort also has several limitations. First, participants are known to have more health-conscious behaviours and better occupational status and education level, compared to the general population. The excluded population (those who did not complete the early-life and adulthood environmental exposures questionnaire) was younger and more often women, in contrast to the study population. Therefore, the results are not necessarily generalisable to the whole population, including other ethnic populations. Secondly, some important exposures for respiratory health could not be taken into account, such as aeroallergens (pollens), indoor environment (dampness) or built environment factors (facilities density). Finally, the authors opted not to use a binary criterion to define asthmatic disease, which seems to be methodologically justified in epidemiological studies [15]; it could however induce a classification bias that may nonetheless lead to confusion with other respiratory diseases, as cough is a nonspecific symptom.

Overall, the INMA and RHINESSA studies provide new insights into the effects of phthalates, air pollution and greenness that are complementary to those of the NutriNet-Santé study, which considered these exposures together in an exposome approach.

Implications for practice and research

These three studies highlight targets for action in primary and secondary prevention. The INMA and RHINESSA studies support the regulation and reduction of phthalates and air pollution. Beyond these specific risk factors, the exposome approach conducted in the NutriNet-Santé study emphasises that risk reduction must involve a multi-interventional approach targeting both early-life risk factors and promotion of a healthy lifestyle in adulthood.

Because environmental exposures are ubiquitous and can be difficult to avoid, it may seem difficult to make an impact in clinical practice on these exposures. However, the role of health professionals is crucial for the prevention and management of environmental exposures. First, a detailed clinical examination of the patient's living environment is necessary to apply the most effective adaptation strategies [16]. In addition, health professionals are in the front line in providing patients with reliable scientific information about the adverse health effects of specific environmental exposures [17]. Finally, health professionals have the opportunity to make their voices heard by health policy makers through the scientific societies to which they may belong [17]. For example, the European Respiratory Society has on several occasions taken a position and called on the European Union to take measures to reduce air pollution levels [17].

All of these points imply initial and ongoing training of health professionals on the health impacts of environmental factors. They also imply the development of scientific knowledge in the field, as there are still many gaps in the quantification of the effects of environmental exposures. The three articles discussed in this journal club allow establishment of research perspectives in the quantification of exposures to phthalates, air pollution and green spaces from early life to adulthood in the context of lung health, as well as the application of the exposome approach in studying multifaceted respiratory diseases. These articles also put into perspective the complexity of the associations between environmental exposures and different methods that can be used to address them.