Introduction
Systemic sclerosis (SSc) is a systemic disease characterized by autoimmune features, and endothelial and fibroblast dysfunctions, resulting in vasculopathy and tissue fibrosis. Interstitial lung disease (ILD) is common in SSc. In a recent nationwide cohort study in Norway, high-resolution computed tomography (HRCT) showed that half the patients had ILD [
1]. ILD has a major impact on the morbidity and mortality of SSc patients, one third of whom die from pulmonary fibrosis [
2]. The course of SSc-associated ILD is heterogeneous [
3]. Extensive lung parenchyma involvement according to the Goh staging algorithm [
4] and short-term functional decline [
5,
6] have been described as predictors of poorer survival, but the factors underlying the prognostic heterogeneity between patients are not fully understood. Nevertheless, ethnic, immunological and phenotypic characteristics of SSc, such as Afro-Caribbean origin, anti-topoisomerase I antibodies, esophageal diameter, reflux/dysphagia symptoms, modified Rodnan skin score, diffuse cutaneous phenotype (dcSSc) and being male [
3,
7‐
9], have been shown to be associated with ILD severity and progression.
Air pollution has been implicated in idiopathic pulmonary fibrosis (IPF) in a number of studies. The incidence of IPF was associated with levels of exposure to nitrogen dioxide (NO
2) and particulate matter of up to 2.5 µm in diameter (PM
2.5) [
10,
11]. Disease severity has been linked to exposure to particulate matter of up to 10 µm or up to 2.5 µm in diameter (PM
10 and PM
2.5) [
12]. Disease exacerbations were linked to exposure to ozone (O
3), NO
2, PM
10, and PM
2.5 [
13‐
16], and functional decline with exposure to PM
10 [
14]
. Mortality was associated with exposure to PM
10, PM
2.5 and NO
2 [
12,
14,
17‐
19]. A role for air pollution in SSc was first suggested by a British study reporting a higher prevalence of SSc in the London region, particularly in boroughs close to airports, than in the West Midlands [
20]. More recently, an Italian study on 88 SSc patients found that benzene exposure was positively correlated with skin score and inversely correlated with the diffusion of carbon monoxide in the lung (DLCO) [
21]. However, to our knowledge, no larger-scale study has evaluated the impact of air pollution on SSc-associated ILD.
We conducted a retrospective study to evaluate the contribution of the principal air pollutants (PM
10, PM
2.5, NO
2, and O
3) to the natural course of SSc-associated ILD. The primary objective was to determine the association between air pollution exposure and disease severity at diagnosis according to the Goh staging algorithm [
4]. The secondary objective was to evaluate the impact of air pollution on disease progression.
Discussion
We investigated the association between the severity of SSc-associated ILD and chronic exposure to PM2.5, PM10, NO2 and O3 in a cohort of SSc patients seen at two hospitals in the Paris area. We observed an association between long-term exposure to O3 and ILD severity at diagnosis, evaluated with the Goh staging algorithm, and according to extension on HRCT, TLC and DLCO. This association was independent of the principal factors associated with the severity of SSc-associated ILD and was confirmed in two-pollutant models. We also found an association between O3 exposure and progression at 24 months. We found no association between exposure to other pollutants and severity at diagnosis and progression.
Ozone is a secondary pollutant generated principally by the photochemical reaction of nitric oxides and oxygen molecules in the atmosphere. It has detrimental effects at concentrations only three to four times higher than natural background levels [
32]. Episodes of high O
3 concentration occur in urbanized areas during periods of sunny anticyclonic weather in the summer months. O
3 is a highly reactive gas, and a powerful oxidant. Epidemiological studies have shown chronic O
3 exposure to be associated with the risk of death from respiratory causes [
33], and the incidence and mortality of acute respiratory distress syndrome [
34,
35]. Animal models and lung autopsy study have revealed the presence of chronic epithelial changes, including fibrosis, in subjects chronically exposed to high O
3 concentrations [
36,
37]. In its Integrated Science Assessment for Ozone, the United States Environmental Protection Agency estimated that there is a “causal relationship” and a “likely causal relationship” between short-term and long-term O
3 exposure, respectively, and respiratory effects [
38]. In IPF, the onset of an acute exacerbation has been shown to be associated with an increase in O
3 exposure within the preceding 6 weeks; however, no association between long-term O
3 exposure and IPF severity has ever been reported [
13]. Nevertheless, long-term exposure to O
3 has been shown to be positively associated with serum IL-4 levels in IPF patients, and tends to be associated with osteopontin levels, two mediators implicated in fibrosis [
39].
The role of air pollution in autoimmune diseases has been studied essentially in rheumatoid arthritis (RA). Particulate matter, such as diesel emission particles, is thought to induce the citrullination of lung proteins and the development of inducible bronchus-associated lymphoid tissue (iBALT), leading to the production of pathogenic anti-citrullinated protein antibodies (ACPA) [
40,
41]. iBALT hyperplasia and the activation of T-cells contained in pulmonary lymph nodes have been observed in animal models exposed to O
3 [
42,
43]. In SSc patients, exposure to O
3 may trigger the development of iBALT-inducing pathogenic autoantibodies. Lung oxidant/antioxidant equilibrium is disturbed at high levels of O
3 exposure, or in situations in which the lung lining fluid antioxidant power is compromised. The reaction of O
3 with substrates present in the lung lining fluid compartment then generates secondary oxidation products and inflammation [
32,
44]. Reactive oxygen species (ROS) have profibrogenic effects on fibroblasts and induce the release of profibrotic mediators, such as transforming growth factor-β 1 (TGF β 1) [
45]. High levels of ROS, produced by the NADPH oxidase system, have been implicated in the pathophysiology of SSc [
46‐
48]. Scleroderma fibroblasts cannot respond to oxidative stress and they mount an inadequate antioxidant response [
46].
Borghini et al. reported that exposure to benzene was inversely correlated with DLCO and positively correlated with Rodnan skin score in SSc patients, whereas they found no association with PM
10 exposure [
21]. Benzene is mostly emitted during wood heating in human homes and in the transport sector and contribute to the formation of O
3 through reactions involving nitrogen oxides (NOx) and solar radiation. Recently, Goobie et al. reported the association of PM
2.5 exposure with lung function at baseline and mortality among patients with fibrotic ILDs [
49]. Homogeneity of particulate matter exposure among our patients could have limited the evaluation of their impact on ILD severity and progression. To our knowledge, ours is the first study to evaluate the effect of O
3 exposure in SSc patients.
Our work has several limitations. First, due to the rarity of SSc and the two-center design of the study, the number of patients included was small for the purpose to detect correlations. Most of the patients were living in the same region, limiting the variability of exposure. However, the use of the CHIMERE model increased the accuracy of exposure estimates, making it possible to detect smaller differences in exposure than would have been possible with the use of concentration data from air quality stations. A limitation inherent to the study design is the estimation of personal exposure at residential addresses, while exposures take place in multiple locations. Assuming that the error in the estimates is random, it would likely bias any association to zero, suggesting that the true magnitude of the effect may be greater than measured [
50]. Mean annual O
3 exposure were considered in our analysis, while the maximum O
3 concentrations are reached during the daytime period in the summer months. Thus, the association of high O
3 levels with respiratory effects may have been underestimated. Trends in the exposure to air pollutants over time may be a source of confounding. However, concentrations of particulate matters and NO
2 have fallen over the years, whereas concentrations of O
3 have increased in our study population, whereas the trend over the years regarding ILD in SSc patients may be assumed toward an earlier diagnosis through HRCT and a better outcome. Moreover, our results were consistent after adjustment for the year of ILD diagnosis. The study was retrospective. As a result, evaluation was not standardized and many PFT data were missing at 12 and 24 months, limiting evaluations of the effects of pollution on progression in our study. Inclusion period (2006–2019) limited follow-up time and the interpretation of the results of survival data. Most of the patients included were followed in the French referral center for SSc (Internal medicine department, Cochin Hospital) or the competence center for ILDs (Respiratory medicine department, Avicenne Hospital), and may therefore not represent a general SSc population. Last but not least, our study’s primary objective was to determine the association between SSc-associated ILD severity at diagnosis but did not consider the incidence of ILD in SSc patients. We did not study the effect of exposure to pollution on extrapulmonary SSc manifestations. Therefore, the role of air pollution exposure on ILD occurrence and other organ involvements in SSc patients remains to be determined.
Despite the retrospective nature of this study, HRCT characteristics at ILD diagnosis, reviewed by expert radiologists, were available for all the patients included, together with PFT parameters, allowing an accurate evaluation of ILD at diagnosis. Our study showed consistently significant associations with O3, however, supportive evidence from future studies in various geographic areas and animal models describing pathophysiological pathways implicated will be necessary to strengthen the arguments for causality.
In conclusion, this study is the first to assess the impact of air pollution on SSc-associated ILD. It reveals an association between O3 exposure and ILD severity at diagnosis and progression at 24 months, that is independent of the principal factors associated with disease severity and progression. The identification of this preventable risk factor could lead to avoidance measures, particularly during periods of high O3 levels in warm weather. A prospective larger-scale multicenter study with a standardized evaluation of progression and prolonged follow-up is required, to confirm our results and to assess the effect of air pollution exposure on SSc-associated ILD incidence and outcome.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.