Skip to main content

Long-term exposure to a mixture of industrial SO2, NO2, and PM2.5 and anti-citrullinated protein antibody positivity

Abstract

Background

Studies of associations between industrial air emissions and rheumatic diseases, or diseases-related serological biomarkers, are few. Moreover, previous evaluations typically studied individual (not mixed) emissions. We investigated associations between individual and combined exposures to industrial sulfur dioxide (SO2), nitrogen dioxide (NO2), and fine particles matter (PM2.5) on anti-citrullinated protein antibodies (ACPA), a characteristic biomarker for rheumatoid arthritis (RA).

Methods

Serum ACPA was determined for 7600 randomly selected CARTaGENE general population subjects in Quebec, Canada. Industrial SO2, NO2, and PM2.5 concentrations, estimated by the California Puff (CALPUFF) atmospheric dispersion model, were assigned based on residential postal codes at the time of sera collection. Single-exposure logistic regressions were performed for ACPA positivity defined by 20 U/ml, 40 U/ml, and 60 U/ml thresholds, adjusting for age, sex, French Canadian origin, smoking, and family income. Associations between regional overall PM2.5 exposure and ACPA positivity were also investigated. The associations between the combined three industrial exposures and the ACPA positivity were assessed by weighted quantile sum (WQS) regressions.

Results

Significant associations between individual industrial exposures and ACPA positivity defined by the 20 U/ml threshold were seen with single-exposure logistic regression models, for industrial emissions of PM2.5 (odds ratio, OR = 1.19, 95% confidence intervals, CI: 1.04–1.36) and SO2 (OR = 1.03, 95% CI: 1.00–1.06), without clear associations for NO2 (OR = 1.01, 95% CI: 0.86–1.17). Similar findings were seen for the 40 U/ml threshold, although at 60 U/ml, the results were very imprecise. The WQS model demonstrated a positive relationship between combined industrial exposures and ACPA positivity (OR = 1.36, 95% CI: 1.10–1.69 at 20 U/ml) and suggested that industrial PM2.5 may have a closer association with ACPA positivity than the other exposures. Again, similar findings were seen with the 40 U/ml threshold, though 60 U/ml results were imprecise. No clear association between ACPA and regional overall PM2.5 exposure was seen.

Conclusions

We noted positive associations between ACPA and industrial emissions of PM2.5 and SO2. Industrial PM2.5 exposure may play a particularly important role in this regard.

Peer Review reports

Introduction

Air pollution is a major risk factor for cardiorespiratory and chronic airway diseases [1,2,3]. By contrast, studies of air pollution and rheumatic diseases and/or their serologic biomarkers are relatively few, and conclusions from these limited studies are inconsistent [4]. Laboratory studies have shown that ambient air pollutants inhaled and deposited in the lungs can increase airway inflammation [5, 6], triggering systemic autoimmune responses (and possibly facilitating the development of autoimmune rheumatic disease) [7]. However, positive associations between air pollution exposure and autoimmune responses and/or rheumatic disease onset have not always been observed in observational studies [8].

Rheumatoid arthritis (RA) is the most common world-wide chronic inflammatory disease and causes great disability [9]. Anti-citrullinated protein antibodies (ACPA) are a characteristic finding in RA, often predating clinical manifestations of the disease by years [10]. We previously reported that exposure to industrial air emissions, e.g. sulfur dioxide (SO2) and fine particles matter (PM2.5), was associated with increased probability of ACPA positivity in a general population sample [11]. However, in that study a rough proxy of exposure (i.e., distance to major industrial emitters) was used and the number of positive ACPA cases was relatively small.

As well, people are exposed to mixtures of multiple pollutants, yet the joint effects of different air pollutants have not been previously considered in studies of air pollution and rheumatic autoimmune diseases and/or serologic biomarkers. Concentrations of regional ambient air pollutants, and especially industrial air pollutants, are usually correlated in space [12], since these pollutants are often derived from the same sources (e.g. road traffic and factories). Hence, special analytic approaches that can effectively address collinearity should be used for exploring the associations between inter-correlated exposures and the outcome of interest [13].

Given the paucity of studies on individual air pollutant exposures and rheumatic diseases, and the absence of prior evaluations of rheumatic-related antibodies and multi-pollutant mixtures, we expanded our previous analyses within a population-based cohort in Quebec, Canada [11], to investigate associations between exposures to three industrial air pollutants (i.e. SO2, nitrogen dioxide - NO2, and PM2.5) and ACPA positivity. In this new effort, we doubled the sample size, used more accurate pollutant estimates derived from a three-dimensional atmospheric model (California Puff, CALPUFF), and evaluated multiple thresholds for defining ACPA positivity. Moreover, a weighted quantile sum (WQS) regression model [14] was used to detect the joint effect of the multi-pollutant exposures on ACPA positivity.

Methods

Study population and sera samples

Our analyses were based on the CARTaGENE cohort (www.cartagene.qc.ca), which is composed of 43,000 general population subjects aged between 40 to 69 years old, with residential history equal to or longer than 5 years in Quebec, Canada. CARTaGENE is part of the Canadian Partnership for Tomorrow Project, a prospective cohort study created as a population-health research platform for assessing the effect of genetics, behaviour, family health history and environment (among other factors) on chronic diseases [15]. Participants in the CARTaGENE cohort were randomly selected from the provincial health insurance database and invited to participate. At baseline, CARTaGENE data were generated at enrolment and included a wide range of health-related variables such as demographics, medical history, lifestyle factors like smoking, and self-reported RA (past diagnosed by physicians) [16], and baseline serum samples were biobanked. The original smoking variable in the CARTaGENE baseline dataset had four categories that were daily, past, occasional, and never smoking. We incorporated the past smoking category into the occasional smoking category because only 4.0% of subjects were past smokers. Thus, in our analyses, individuals reporting anything other than daily or never smoking were categorized as occasional/past smokers.

For the current study, we selected a random sample of 7600 individuals from the first CARTaGENE recruitment wave (enrolled over 2009–2010). This sample size is twice as large as that of our previous study [11]. Biobanked serum samples were assessed for ACPA positivity by chemiluminescence immunoassay (CCP3.0; Inova Diagnostics, San Diego, CA, USA) at the Mitogen Advanced Diagnostics Lab in Calgary, Alberta. ACPA positivity was defined initially on the basis of test results ≥20 U/ml [17]. In sensitivity analyses, two other thresholds were also used, to classify all positive ACPA outcomes as weak (20–39 U/ml), moderate (40–59 U/ml), and strong (≥60 U/ml) positive titres [18].

Air pollution exposures

CALPUFF is an advanced dispersion modeling system, which can simulate the effects of spatiotemporally varying meteorological conditions on transport, transformation, and dissipation of air pollutants [19]. The modeling system consists of three major components namely CALMET (a three-dimensional meteorological model), CALPUFF (an air quality dispersion model), and CALPOST (a post-processing package). The CALPUFF system is recommended by the United States Environmental Protection Agency to assess long-range tracking of air pollutants and has been extensively used to map regional SO2, NO2, and particulate matter concentrations in Canada, the United States, and other countries [20,21,22,23,24,25]. In this study, using industrial emissions reported to the National Pollutant Release Inventory [26], industrial SO2, NO2, and PM2.5 annual average levels for 2005–2010 were modeled by the CALPUFF at the locations of Quebec’s six-digit postal codes and then were assigned to each subject based on his/her postal code at the time of CARTaGENE enrollment (when blood samples were taken). Please see the paper of Buteau et al. (2020) [27] for details of using the CALPUFF modelling system to estimate industrial SO2, NO2, and PM2.5 annual average concentrations.

The annual average regional (overall but not only industrial) PM2.5 concentration estimates were retrieved from the Atmospheric Composition Analysis Group at Dalhousie University. The PM2.5 concentrations were first estimated at the 10 km resolution using the GEOS (Goddard Earth Observing System) chemical-transport-model and the satellite-derived aerosol optical depth data [28]. These coarse gridded PM2.5 products were further resampled to the 1 km spatial resolution by a geographical weighted regression model and additional covariates, e.g. elevation, vegetation index, and distance to urban areas [29]. Similar to the industrial SO2, NO2, and PM2.5 exposures, the average regional (all-sector) PM2.5 estimates for 2005–2010 extracted from the above raster dataset were assigned to all the participants based on their six-digit postal codes at the time of the cohort enrollment. Since all participants entered into the CARTaGENE cohort during 2009 to 2010 and had residential history in Quebec equal to or longer than 5 years, we were assured that the participants have been in Quebec from at least 2005. Thus, as in our previous study [30], we selected the exposure time window of 2005–2010 to ensure that subjects’ assigned long-term air pollution exposures represented their actual exposures.

Standard logistic regression models

We first used three separate single-exposure standard logistic regression models, adjusting for age, sex, ancestry, smoking, and family income (see Table 1 for the detailed categories of the covariates), to detect the associations between individual industrial SO2, NO2, and PM2.5 exposures and ACPA positivity (defined by the 20 U/ml threshold). These covariates were chosen as they may be potential effect modifiers (e.g. sex) or confounders (e.g. age, French Canadian ancestry, family income, and smoking) of relationships between variations in air pollution and serologic positivity [11]. The single-exposure logistic regression adjusting for the same covariates was also conducted for regional overall PM2.5 exposure, to examine whether the same air pollutant from different (i.e. regional overall vs. industrial) emission sources would produce different effects on ACPA positivity. Next, we increased the threshold of defining ACPA positivity to 40 U/ml and 60 U/ml, and used the above single-exposure logistic regression models in sensitivity analyses. We did not use multi-exposure logistic regressions to investigate the associations of combined exposures to industrial SO2, NO2, and PM2.5, because concentrations of the three industrial air pollutants are closely correlated in space (see the Results section for specific correlation coefficients). To see whether air pollution exposures have different effects on ACPA and RA, we also used the standard logistic regression models, adjusting for age, sex, ancestry, smoking, and family income, to detect the associations between RA and individual industrial SO2, NO2, and PM2.5 and regional overall PM2.5 exposures.

Table 1 Baseline characteristics of the subjects and distributions of pollutants according to antibody outcomes

The WQS regression models

The joint association of the three highly correlated industrial air pollutants with ACPA positivity was explored by the WQS regression method and quantitatively assessed by a WQS index [14]. The WQS approach supposes that all the studied exposures have the same direction (positive or negative) effects on the disease outcome. Magnitudes of the individual effects of different exposures are quantified by a set of weights. Each of the weights is constrained to be between 0 and 1, and all of the weights are summed to 1. The weights were multiplied by the scored quartiles of the individual exposures, and then were accumulated to obtain the WQS index.

To calculate the weights, we first performed natural logarithm transformations on the three exposure variables to ensure each had similar scales. Then, we divided the sample into a training and a validation datasets using a split proportion of 4:6. This proportion was adopted by the previous WQS studies [14, 31] because leaving more subjects in the validation dataset tends to increase robustness for calculating the significance of the WQS index [14]. A total of B = 100 bootstrap samples were generated from the training dataset to estimate the unknown weight wi (i denoting one of the industrial air pollutants) by maximizing the likelihood of the weighted index function:

$$ {\left.g\left(\mu \right)={\beta}_0+{\beta}_1\left({\sum}_{i=1}^3{w}_i{q}_i\right)+{\beta}^Tz\right|}_b\kern0.5em \left(b=1,2,\dots, 100\right) $$
(1)

where g(·) is a logit link function for the binary outcome of a positive (or negative) ACPA, z denotes a vector of potential confounders or effect modifiers (i.e. age, sex, French Canadian ancestry, smoking, and family income), β is the coefficient vector of the covariates, β0 is the intercept, q represents a quartile of the logarithmically transformed exposure. The term \( {\sum}_{i=1}^3{w}_i{q}_i \) represents the weighted index and β1 is its regression coefficient. Let \( WQS={\sum}_{i=1}^3{w}_i{q}_i \), and thus the eq. 1 can be simplified as.

$$ g\left(\mu \right)={\beta}_0+{\beta}_1 WQS+{\beta}^Tz $$
(2)

The odds ratio (OR) associated with a quartile increase in all of the three logarithmically transformed exposures (i.e. the WQS index) is equal to exponentiated β1.

The specific WQS regression was implemented using the “gWQS” package [32] in the R statistical computing environment. Similar to the single-exposure logistic regressions, the WQS regressions were conducted three times for positive ACPA outcomes defined by the three thresholds (i.e. 20 U/ml or higher, 40 U/ml or higher, and 60 U/ml or higher).

RA affects less than 1% of the general population of Quebec [33]. After splitting our sample into a training and a validation datasets, we did not have enough RA cases in either dataset for a reliable fitting or validation. Thus, we did not use WQS regression to detect the relationship between combined industrial exposures and RA in this study.

Results

In the total 7600 subjects the mean age at cohort entry was 54.1 years (standard deviation, SD =7.7 years) and 3859 (50.8%) were female. Approximately two-third (67.3%) of the subjects were French Canadians. Over 40 % (N = 3053, 40.2%) of the subjects were never smokers, 1020 (13.4%) were daily smokers, 3492 (45.9%) were occasional/past smokers, and the remainder (N = 26) had missing smoking data. Only 9.3% of the population subjects lived below the lowest household income level (i.e. < 25,000 Canadian dollars per year) while 11.5% belonged to the highest level for income (i.e. ≥150,000 Canadian dollars per year). Detailed comparisons among the strong, moderate, and weak ACPA positive and negative subjects are presented in Table 1. A total of 201 subjects in our sample reported physician-diagnosed RA when they entered the cohort, and 37 individuals had both RA and positive ACPA. Furthermore, 24 of the 37 individuals had ACPA ≥60 U/ml.

The interquartile ranges of the logarithmically transformed industrial SO2, NO2, and PM2.5 exposures were 1.34 ppb, 1.04 ppb, and 1.58 μg/m3, respectively. Pearson’s correlations coefficients (r) indicated that besides a moderate correlation between industrial PM2.5 and regional overall PM2.5 concentrations (r = − 0.13, p < 0.001, 95% confidence intervals, CI -0.16 – − 0.11), industrial PM2.5 levels were strongly correlated to those of industrial SO2 (r = 0.96, p < 0.001, 95% CI: 0.96–0.97) and moderately to NO2 (r = 0.19, p < 0.001, 95% CI: 0.17–0.21); the concentration of SO2 was also moderately correlated with NO2 (r = 0.35, p < 0.001, 95% CI: 0.33–0.37).

As presented in Table 2, clearly positive associations between industrial SO2 (OR: 1.03, 95% CI: 1.00–1.06) and PM2.5 (OR: 1.19, 95% CI: 1.04–1.36) exposures and ACPA positivity were observed from the standard single-exposure regression analyses, when the ACPA positivity was defined by the 20 U/ml threshold. With the threshold increased to 40 U/ml, the positive associations of industrial SO2 (OR: 1.03, 95% CI: 1.00–1.07) and PM2.5 (OR: 1.21, 95% CI: 1.02–1.42) exposures with ACPA positivity were similar. However, when the ACPA threshold was further increased to 60 U/ml, the point estimates were similar but the 95% CIs became wider due to very low numbers of cases (industrial SO2 OR: 1.03, 95% CI: 0.98–1.08 and industrial PM2.5 OR: 1.17, 95% CIs: 0.92–1.48). Industrial NO2 and regional overall PM2.5 exposures were not clearly associated with ACPA positivity, regardless of the thresholds used to define positivity (Table 2). Positive ACPA was more common in subjects of older age (as is expected, given that both RA and ACPA are more common in older individuals) [34]. Due to low power, we did not see a clear relationship between RA and any air pollutant exposure (see Table S1). A few previous studies (e.g. [35,36,37,38]) have found that smoking increased the risk of developing ACPA-positive RA, but we did not find a clear relationship of smoking with either ACPA positivity or RA (Table S2).

Table 2 Adjusted OR (95% CIs) from the single-pollutant logistic regression models for ACPA positivity

The WQS index (i.e. the mixture of the three industrial air emissions) was significantly correlated with ACPA positivity defined by the 20 U/ml threshold. Specifically, an interquartile increase in the WQS index led to an increase of 1.36 (95% CI: 1.10–1.69) in the odds of ACPA positivity. With the positivity threshold increased to 40 U/ml, the positive association between the combined logarithmically transformed exposure of the three industrial air pollutants and ACPA positivity was still apparent (OR = 1.43, 95% CI: 1.05–1.96). When the ACPA positivity was defined by a higher threshold of 60 U/ml, due to low numbers of cases, the association between the WQS index and ACPA positivity became less clear (OR = 1.33, 95% CI: 0.85–2.10). Regardless of the threshold for ACPA positivity, industrial PM2.5 was always the most heavily weighted while the industrial NO2 was the most lightly weighted in the index (Table 3).

Table 3 Adjusted OR (95% CIs) from the weighted quantile sum (WQS) regressions for ACPA positivity

Discussion

Exposure to ambient air pollutants may induce pulmonary oxidative stress and inflammation [39, 40] and consequently trigger autoimmune responses which could favor the development of RA and related diseases [7, 41, 42]. However, results of early epidemiologic studies have not always supported this hypothesis. Although, Hart et al. (2013) [43] found that exposure to NO2 from road traffic is likely to increase risk of RA incidence using a Swedish general population cohort, positive associations between RA incidence and exposure to NO2 and PM2.5 were not observed by De Roos et al. (2014) [8] in British Columbia, Canada. Gan et al. (2013) [9] also did not find conclusive associations between ambient particulate matter exposure and RA-related antibodies in first-degree relatives of RA patients in the United States, but a clear association between ambient NO2 and RA was found by Chang et al. (2016) [44] in Taiwan. Due in part to low power, we did not see a clear relationship between RA and any air pollutant exposure. Linking baseline CARTaGENE data with administrative data could be a way to generate follow-up data, which may allow us to obtain information on new cases of RA within CARTaGENE, to better study the relationships between RA and air pollution exposures in the future [33].

Several previous studies using cohorts from Europe reported that smoking could increase the risk of ACPA-positive RA while a conclusive association between smoking and ACPA-negative RA was not observed [36,37,38]. In our study, only 4.7% of ACPA-positive subjects reported a physician diagnosis of RA before they entered the cohort. Thus, failure to find associations between smoking and ACPA positivity dose not necessarily contradict the findings from Europe [35,36,37,38] and is consistent with our previous finding in Quebec [11].

In our mixed-pollutant analyses, regardless of the threshold for ACPA positivity, industrial PM2.5 appeared to be the most influential exposure, while exposure to industrial NO2 was the least influential. In an earlier study using CARTaGENE data, we found that exposures to industrial SO2 and PM2.5 were associated with ACPA positivity, but no clear associations were seen with industrial NO2 and ambient PM2.5 [11]. In the current study, we reinforced these findings with twice the sample size, and more accurate estimates of exposure to industrial air emission (since the prior study used simple distance to major industrial emitters). Most importantly, the use of the WQS model allowed us to assess the combined association of all three industrial emissions and to quantify different contributions of the individual emissions on ACPA positivity, which is more representative of how people are always exposed to multiple, correlated air pollutants. We found similar associations between ACPA and industrial PM2.5 and SO2, at both low and medium titres, although limiting positivity to very higher titres led to imprecise results.

Surface chemistry of industrial ambient particulate matter is likely to be more toxic than that of regional overall ambient particulate matter [45], which may explain why ACPA positivity was associated with industrial PM2.5 exposure but not ambient PM2.5 exposure. Additionally, we found that the industrial PM2.5 concentration was negatively correlated with the regional overall PM2.5 concentration. In other words, individuals exposed to higher ambient PM2.5 levels are less likely to be exposed to high industrial PM2.5 concentrations. This might be explained by the fact that industrial emitters of PM2.5 tend to be located away from high-traffic areas (since motor vehicles account for the majority of ambient PM2.5 levels to which people are exposed) [46].

When studying the joint effects of multiple air pollutant exposures it may be preferable to use the WQS approach rather than standard logistic regression models, in order to avoid the problem of collinearity. However, the WQS method has a critical restriction, in that if the studied exposures have effects on the disease outcome that differ in direction (i.e. positive versus negative), the model will not converge. The Bayesian kernel machine regression (BKMR) [47] is an alternative method to study the combined effects of multiple correlated exposures on binary disease outcomes, without this critical restriction. However, the computing time of fitting a BKMR model increases exponentially with an increase in the number of subjects. By contrast, the WQS model is much more efficient for a large sample, which prompted us to choose the WQS and not the BKMR approach in this study.

Participants in the CARTaGENE cohort were all aged between 40 to 69 years old. That is a potential limitation of this study, since younger individuals may be more susceptible to the adverse health effects from air pollution [42]. However, people in the 40 to 69 age group are less mobile than younger (e.g. college age) ones. This may be beneficial in terms of reducing errors when postal codes at a single point in time are used to assign exposure information like in this study (where highly mobile populations may be subject to more misclassification of exposure).

In this study, the average level of industrial NO2 in the population under study was very low, and variation in the industrial NO2 concentration across Quebec is small, which may be a reason why we failed to observe a clear association between industrial NO2 exposure and ACPA positivity. Thus, additional studies including younger populations and conducted in higher industrial NO2 regions may help reinforce or refute the current findings. Besides air pollution exposures, occupational dust exposures (e.g. asbestos, silica, and carbon nanoparticles) are also likely to be associated with ACPA positivity and RA [48, 49]. Thus, occupation may need to be added as a covariate in the next studies regarding industrial air pollution exposures and ACPA. In addition, further study of air pollution and RA onset may be informative, particularly if more sophisticated approaches (such as WQS regression or BKMR) are employed. Another future direction may be to examine RA-related manifestations, such as pulmonary disease, and air pollution [50].

Conclusions

Using a larger sample, more accurate exposure estimates, and more detailed positivity definition than in the past, this study reinforced our previous findings that exposures to industrial sources of SO2 and PM2.5 tend to increase the probability of ACPA positivity. No clear association between ACPA positivity and industrial NO2 or regional overall PM2.5 exposure was detected. The use of the WQS approach allowed us to produce new findings concerning positive associations between mixtures of the industrial air pollutants, and suggested that the effects of industrial PM2.5 exposure on ACPA positivity may be more important than that of industrial SO2 exposure.

Availability of data and materials

The datasets used in the current study are available from the corresponding author on reasonable request.

References

  1. Brunekreef B, Holgate ST. Air pollution and health. Lancet. 2002;360(9341):1233–42.

    CAS  Google Scholar 

  2. Dominici F, Peng R, Bell ML, Pham L, McDermott A, Zeger SL, Samet JM. Fine particulate air pollution and hospital admission for cardiovascular and respiratory diseases. JAMA. 2006;295(10):1127–34.

    CAS  Google Scholar 

  3. Mills N, Donaldson K, Hadoke PW, Boon NA, MacNee W, Cassee FR, et al. Adverse cardiovascular effects of air pollution. Nat Clin Pract Cardiovasc Med. 2009;6:36–44.

    CAS  Google Scholar 

  4. Sun G, Hazlewood G, Bernatsky S, Kaplan GG, Eksteen B, Barnabe C. Association between air pollution and the development of rheumatic disease: a systematic review. Int J Rheumat. 2016;2016:5356307.

    Google Scholar 

  5. Card JW, Zeldin DC, Bonner JC, Nestmann ER. Pulmonary applications and toxicity of engineered nanoparticles. Am J Physiol Lung Cell Mol Physiol. 2008;295(3):L400–11.

    CAS  Google Scholar 

  6. Romieu I, Castro-Giner F, Kunzli N, Sunyer J. Air pollution, oxidative stress and dietary supplementation: a review. Eur Respir J. 2008;31:179–97.

    CAS  Google Scholar 

  7. Farhat SCL, Silva CA, Orione MAM, Campos LMA, Sallum AME, Braga ALF. Air pollution in autoimmune rheumatic diseases: a review. Autoimmun Rev. 2011;11(1):14–21.

    CAS  Google Scholar 

  8. Di D, Zhang L, Wu X, Leng R. Long-term exposure to outdoor air pollution and the risk of development of rheumatoid arthritis: a systematic review and meta-analysis. Semin Arthritis Rheum. 2020;50:266–75.

    Google Scholar 

  9. Cross M, Smith E, Hoy D, Carmona L, Wolfe F, Vos T, et al. The global burden of rheumatoid arthritis: estimates from the global burden of disease 2010 study. Ann Rheum Dis. 2014;73:1316–22.

    Google Scholar 

  10. Puszczewicz M, Iwaszkiewicz C. Role of anti-citrullinated protein antibodies in diagnosis and prognosis of rheumatoid arthritis. Arch Med Sci. 2011;7(2):189–94.

    CAS  Google Scholar 

  11. Bernatsky S, Smargiassi A, Joseph L, Awadalla P, Colmegna I, Hudson M, et al. Industrial air emissions, and proximity to major industrial emitters, are associated with anti-citrullinated protein antibodies. Environ Res. 2017;157:60–3.

    CAS  Google Scholar 

  12. De Roos AJ, Koehoorn M, Tamburic L, Davies HW, Brauer M. Proximity to traffic, ambient air pollution, and community noise in relation to incident rheumatoid arthritis. Environ Health Perspect. 2014;122:1075–80.

    Google Scholar 

  13. Bellavia A, James-Todd T, Williams PL. Approaches for incorporating environmental mixtures as mediators in mediation analysis. Environ Int. 2019;123:368–74.

    Google Scholar 

  14. Carrico C, Gennings C, Wheeler D, Factor-Litvak P. Characterization of a weighted quantile sum regression for highly correlated data in a risk analysis setting. J Agric Biol Environ Stat. 2015;20(1):100–20.

    Google Scholar 

  15. Dummer TJB, Awadalla P, Boleau C, Craig C, Fortier I, Goel V, et al. The Canadian Partnership for Tomorrow Project: a pan-Canadian platform for researchon chronic disease prevention. Can Med Assoc J. 2018;190:710–7.

    Google Scholar 

  16. Awadalla P, Boileau C, Payette Y, Idaghdour Y, Goulet JP, Knoppers B, et al. Cohort profile of the CARTaGENE study: Quebec’s population-based biobank for public health and personalized genomics. Int J Epidemiol. 2013;42(15):1285–99.

    Google Scholar 

  17. Shafrin J, Tebeka MG, Price K, Patel C, Michaud K. The economic burden of ACPA-positive status among patients with rheumatoid arthritis. J Manag Care Spec Pharm. 2018;24(1):4–11.

    Google Scholar 

  18. Quest Diagnostics. Cyclic citrullinated peptide (CCP) antibody (IgG). Test Center. 2019. Available at: http://www.questdiagnostics.com/testcenter/TestDetail.action?ntc=11173. Accessed 24 Feb 2020.

  19. Exponent. CALPUFF Modeling System. 2018. Available at http://www.src.com/. Accessed 24 Feb 2020.

    Google Scholar 

  20. Abdul-Wahab S, Chan K, Ahmadi L, Elkamel A. Impact of geophysical and meteorological conditions on the dispersion of NO2 in Canada. Air Qual Atmosphere Health. 2014;7(2):113–29.

    CAS  Google Scholar 

  21. Abdul-Wahab S, Sappurd A, Al-Damkhi A. Application of California puff (CALPUFF) model: a case study for Oman. Clean Techn Environ Policy. 2011;13(1):177–89.

    CAS  Google Scholar 

  22. Fisher AL, Parsons MC, Roberts SE, Shea PJ, Khan FI, Husain T. Long-term SO2 dispersion modeling over a coastal region. Environ Technol. 2003;24(4):399–409.

    CAS  Google Scholar 

  23. Hao J, Wang L, Shen M, Li L, Hu J. Air quality impacts of power plant emissions in Beijing. Environ Pollut. 2007;147(2):401–8.

    CAS  Google Scholar 

  24. Henderson SB, Burkholder B, Jackson PL, Brauer M, Ichoku C. Use of MODIS products to simplify and evaluate a forest fire plume dispersion model for PM10 exposure assessment. Atmos Environ. 2008;42(36):8524–32.

    CAS  Google Scholar 

  25. Levy JI, Spengler JD, Hlinka D, Sullivan D, Moon D. Using CALPUFF to evaluate the impacts of power plant emissions in Illinois: model sensitivity and implications. Atmos Environ. 2002;36(6):1063–75.

    CAS  Google Scholar 

  26. Government of Canada. Access data from the National Pollutant Release Inventory. 2019. Available at https://www.canada.ca/en/environment-climate-change/services/national-pollutant-release-inventory/tools-resources-data/access.html. Accessed 24 Feb 2020.

    Google Scholar 

  27. Buteau S, Shekarrizfard M, Hatzopoulou M, Gamache P, Liu L, Smargiassi A. Air pollution from industries and asthma onset in childhood: a population-based birth cohort study using dispersion modeling. Environ Res. 2020;185:109180.

    CAS  Google Scholar 

  28. van Donkelaar A, Martin RV, Park RJ. Estimating ground level PM2.5 using aerosol optical depth determined from satellite remote sensing. J Geophysical Res Atmospheres. 2006;111:D21201.

    Google Scholar 

  29. van Donkelarr A, Martin RV, Brauer M, Hsu NC, Kahn RA, Levy RC, et al. Global estimation of fine particulate matter using a combined geophysical-statistical method with information from satellites, models, and monitors. Environ Sci Technol. 2016;50:3762–72.

    Google Scholar 

  30. Zhao N, Smargiassi A, Hudson M, Fritzler MJ, Bernatsky S. Investigating associations between anti-nuclear antibody positivity and combined long-term exposures to NO2, O3, and PM2.5 using a Bayesian kernel machine regression approach. Environ Int. 2020;136:105472.

    CAS  Google Scholar 

  31. Zhang Y, Dong T, Hu W, Wang X, Xu B, Lin Z. Association between exposure to a mixture of phenols, pesticides, and phthalates and obesity: comparison of three statistical models. Environ Int. 2019;123:325–36.

    CAS  Google Scholar 

  32. Renzetti S, Curtin P, Just AC, Bello G, Gennings C. gWQS: Generalized Weighted Quantile Sum Regression. 2019. Available at: https://rdrr.io/cran/gWQS/. Accessed 27 June 2020.

    Google Scholar 

  33. Slim ZF, de Moura CS, Bernatsky S, Rahme E. Identifying rheumatoid arthritis cases with the Quebec health administrative database. J Rheumatol. 2019;46(12):1570–6.

    Google Scholar 

  34. Bernatsky S, Smargiassi A, Johnson M, Kaplan GG, Barnabe C, Svenson L, et al. Fine particulate air pollution, nitrogen dioxide, and systemic auto-immune rheumatic disease in Calgary, Alberta. Environ Res. 2015;140:474–8.

    CAS  Google Scholar 

  35. Cartwright AJ, Quirke A, de Pablo P, Romaguera D, Panico S, Mattiello A, et al. A1.5 smoking is a risk factor for ACPA prior to onset of symptoms of rheumatoid arthritis in a cohort from southern europe. Ann Rheum Dis. 2014;73:A2–3.

    Google Scholar 

  36. Hedström AK, Stawiarz L, Klareskog L, Alfredsson L. Smoking and susceptibility to rheumatoid arthritis in a Swedish population-based case-control study. Eur J Epidemiol. 2018;33:415–23.

    Google Scholar 

  37. Jiang X, Kallerg H, Chen Z, Arlestig L, Rantapaa-Dahlqvist S, Davila S, et al. An immunochip-based interaction study of contrasting interaction effects with smoking in ACPA-positive versus ACPA-negative rheumatoid arthritis. Rheumatology. 2016;55(1):149–55.

    CAS  Google Scholar 

  38. Lundstrom E, Kallberg H, Alfredsson L, Klareskog L, Padyukov L. Gene-environment interaction between the DRB1 shared epitope and smoking in the risk of anti-citrullinated protein antibody-positive rheumatoid arthritis: all alleles are important. Rheumatoid Arthritis. 2009;60(6):1597–603.

    Google Scholar 

  39. Chuang K-J, Chan C-C, Su T-C, Lee C-T, Tang C-S. The effect of urban air pollution on inflammation, oxidative stress, coagulation, and autonomic dysfunction in young adults. Am J Respir Crit Care Med. 2007;176(4):370–6.

    CAS  Google Scholar 

  40. Liu L, Poon R, Chen L, Frescura A-M, Montuschi P, Ciabattoni G, et al. Acute effects of air pollution on pulmonary function, airway inflammation, and oxidative strees in asthmatic children. Environ Health Perspect. 2009;117(4):668–74.

    CAS  Google Scholar 

  41. Bernatsky S, Fournier M, Pineau CA, Clarke AE, Vinet E, Smargiassi A. Associations between ambient fine particulate levels and disease activity in patients with systemic lupus erythematosus (SLE). Environ Health Perspect. 2011;119(1):45–9.

    CAS  Google Scholar 

  42. Zeft AS, Prahalad S, Lefevre S, Clifford B, McNally B, Bohnsack JF, Pope CA III. Juvenile idiopathic arthritis and exposure to fine particulate air pollution. Clin Exp Rheumatol. 2009;27:877–84.

    CAS  Google Scholar 

  43. Hart JE, Kallberg H, Laden F, Bellander T, Costenbader KH, Holmqvist M, et al. Ambient air pollution exposures and risk of rheumatoid arthritis: results from the Swedish EIRA case-control study. Ann Rheum Dis. 2013;72(6):888–94.

    CAS  Google Scholar 

  44. Chang KH, Hsu CC, Muo CH, Hsu CY, Liu HC, Kao CH, et al. Air pollution exposure increases the risk of rheumatoid arthritis: a longitudinal and nationwide study. Environ Int. 2016;94:495–9.

    CAS  Google Scholar 

  45. Oberdörster G. Pulmonary effects of inhaled ultrafine particles. Int Arch Occup Environ Health. 2000;74(1):1–8.

    Google Scholar 

  46. Karagulian F, Belis CA, Dora CFC, Prüss-Ustün AM, Bonjour S, Adair-Rohani H, Amann M. Contributions to cities’ ambient particulate matter (PM): a systematic review of local source contributions at global level. Atmos Environ. 2015;120:475–83.

    CAS  Google Scholar 

  47. Bobb JF, Valeri L, Henn BC, Christiani DC, Wright RO, Mazumdar M, et al. Bayesian kernel machine regression for estimating the health effects of multi-pollutant mixtures. Biostatistics. 2015;16:493–508.

    Google Scholar 

  48. Speck-Hernandez CA, Montoya-Ortiz G. Silicon, a possible link between environmental exposure and autoimmune diseases: the case of rheumatoid arthritis. Arthritis. 2012;2012:604187.

    Google Scholar 

  49. Murphy D, Hutchinson D. Is male rheumatoid arthritis an occupational disease? A review. Open Rheumatol J. 2017;11:88–105.

    Google Scholar 

  50. Huang S, He X, Doyle TJ, Zaccardelli A, Marshall AA, Friedlander MH, et al. Association of rheumatoid arthritis-related autoantibodies with pulmonary function test abnormalities in a rheumatoid arthritis registry. Clin Rheumatol. 2019;38:3401–12.

    Google Scholar 

Download references

Acknowledgements

ACPA assay kits were a generous gift from Inova Diagnostics, San Diego, CA. The authors acknowledge the laboratory technical efforts of Ms. Haiyan Hou and Ms. Natalia Baeza at Mitogen Advanced Diagnostics (Calgary, AB, Canada). The authors wish to thank the two reviewers for their helpful comments and suggestions. This work was funded by the Canadian Institutes of Health Research (CIHR) (PJT-159682).

Author information

Authors and Affiliations

Authors

Contributions

Naizhuo Zhao: data curation, software, methodology, writing. Audrey Smargiassi: data curation, methodology, writing. Marianne Hatzopoulou: methodology, data curation. Ines Colmegna: methodology, writing. Marie Hudson: methodology, writing. Marvin J. Fritzler: data curation, writing. Philip Awadalla: methodology, writing. Sasha Bernatsky: methodology, supervision, project administration, writing. The author(s) read and approved the final manuscript.

Corresponding author

Correspondence to Sasha Bernatsky.

Ethics declarations

Ethics approval and consent to participate

The study was reviewed by the McGill University, faculty of medicine, ethics review committee and was given full approval to conduct (IRB #: A04-M46-12B). In addition, the CARTaGENE scientific review committee, affiliated with the Centre hospitalier universitaire Sainte-Justine (project #582582) provide approval for data and samples to be analyzed.

Consent for publication

Not applicable.

Competing interests

The authors declare they have no actual or potential competing financial interests.

Additional information

Publisher’s Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary information

Additional file 1: Table S1.

Adjusted OR (95% CIs) from the single-pollutant logistic regression models for RA. Table S2. Adjusted OR (95% CIs) from the single-pollutant logistic regression models for ACPA positivity defined by the 20 unit/ml threshold.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Zhao, N., Smargiassi, A., Hatzopoulou, M. et al. Long-term exposure to a mixture of industrial SO2, NO2, and PM2.5 and anti-citrullinated protein antibody positivity. Environ Health 19, 86 (2020). https://doi.org/10.1186/s12940-020-00637-3

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: https://doi.org/10.1186/s12940-020-00637-3

Keywords