TRANSPHORM. Deliverable D3.2.1, type R

Transcription

1 !! TRANSPHORM Transport related Air Pollution and Health impacts!" Integrated Methodologies for Assessing Particulate Matter Collaborative Project, Large-scale Integrating Project SEVENTH FRAMEWORK PROGRAMME ENV Transport related air pollution and health impacts Deliverable D3.2.1, type R "#$%&'$(&%!)%*#%+!,!-'.%&!/0!$1%!&%2'$#/031#.!4%$+%%0!$&'03./&$!&%2'$%5!'64#%0$!-7!'#&!./22($#/0!'05!#08#5%08%!/9!'3$16' Due date of deliverable: project month 24 Actual submission date: project month 23 Start date of project: 1 January 2010 Duration: 48 months Organisation name of lead contractor for this deliverable: Scientist responsible for this deliverable: UU / Swiss TPH (formerly STI) Bert Brunekreef / Nino Kuenzli Revision: [0]

2 PAPER IN PRESS (confirmation of editor see below) in the Journal Seminars in Respiratory and Critical Care Medicine The role of air pollution in adult-onset asthma - a review of the current evidence Benedicte Jacquemin 1, Tamara Schikowski 2, 3, Anne Elie Carsin 4, Anna Hansell 5, Ursula Krämer 6, Jordi Sunyer 4, Nicole Probst-Hensch 2,3, Francine Kauffmann 1, Nino Künzli 2,3 Affiliations 1. CESP Centre for research in Epidemiology and Population health, U1018, Inserm, Villejuif, France and Université Paris Sud 11, UMRS 1018, Villejuif, France 2. Swiss Tropical and Public Health Institute, Basel, Switzerland 3. University of Basel, Basel, Switzerland 4. Centre for Research in Environmental Epidemiology CREAL, Barcelona, Spain, Consortium for Research on Epidemiology and Public Health (CIBERESP), Spain, and Hospital del Mar Research Institute (IMIM), Barcelona, Spain. Universitat Pompeu Fabra, Barcelona, Spain ; 5. Imperial College London, London, United Kingdom 6. Leibniz Research Institute for Environmental Medicine, Düsseldorf, Germany Corresponding Author: Nino Künzli, MD, PhD, Deputy Director Swiss Tropical and Public Health Institute (SwissTPH); Department of Epidemiology and Public Health & Chair for Social and Preventive Medicine Medical School University Basel Mailing address: Swiss Tropical and Public Health Institute, P.O. Box, 4002 Basel, Switzerland; Tel: +41- (0) Fax: +41- (0) Nino.Kuenzli@unibas.ch Abstract The causes of adult-onset asthma are poorly established and the asthmogenic role of air pollution has been investigated primarily in children. This review assesses the current evidence of the association between air pollution and asthma incidence among subjects free of asthma at least until late childhood. Seven publications from five study populations fulfilled the inclusion criteria (1 case-control and 6 cohort studies). All but one used markers of local traffic-related air pollution to characterize long-term exposure. Those studies reported similar associations with traffic-related air pollution. However, protocols, definitions of asthma, and exposure assignment were rather heterogeneous, and three publications relied on the same study, thus, we abstain from meta-analytic summaries. Reported patterns of effect modification (e.g. by sex, atopy, or smoking) were inconsistent. Overall, the role of trafficrelated air pollution in adult-onset asthma is less conclusive than in childhood asthma. Larger studies with more consistent definitions of phenotypes and exposure assessment for local traffic-related pollutants (e.g. ultrafine particles) are needed. Pooling existing cohorts such as in the ongoing European ESCAPE and TRANSPHORM consortia are promising steps. There is, however, a need for large-scale mega-cohorts to investigate these effects in standardized ways, and to identify the most susceptible populations.!!!! =!

3 From: "Lynch, III, Joseph P." To: Cc: Date: :26 Subject: RE: On Behalf of Dr. Joseph Lynch: chapter for issue on Asthma - Due Now October 14, 2011 Dear Dr. Kunezli, Thank you for sending your outstanding manuscript for the issue of Seminars in Respiratory and Critical Care Medicine on Asthma (Drs. Virchow and Barnes, Guest Editors). We are pleased to accept your article. I have forwarded your manuscript to the publisher. You will receive galley proofs prior to publication. Thank you for you diligence and hard work! Sincerely, Joseph P. Lynch, III, M.D. Editor-in-Chief Holt and Jo Hickman Endowed Chair of Advanced Lung Diseases and Lung Transplantation Professor of Clinical Medicine, Step VII Associate Chief, Division of Pulmonary, Critical Care Medicine, Allergy, and Clinical Immunology David Geffen School of Medicine at UCLA! ;!

4 A copy of the accepted manuscript follows this page.! E!

5 The role of air pollution in adult-onset asthma - a review of the current evidence Benedicte Jacquemin 1, Tamara Schikowski 2, 3, Anne Elie Carsin 4, Anna Hansell 5, Ursula Krämer 6, Jordi Sunyer 4, Nicole Probst-Hensch 2,3, Francine Kauffmann 1, Nino Künzli 2,3 Affiliations 1. CESP Centre for research in Epidemiology and Population health, U1018, Inserm, Villejuif, France and Université Paris Sud 11, UMRS 1018, Villejuif, France 2. Swiss Tropical and Public Health Institute, Basel, Switzerland 3. University of Basel, Basel, Switzerland 4. Centre for Research in Environmental Epidemiology CREAL, Barcelona, Spain, Consortium for Research on Epidemiology and Public Health (CIBERESP), Spain, and Hospital del Mar Research Institute (IMIM), Barcelona, Spain. Universitat Pompeu Fabra, Barcelona, Spain ; 5. Imperial College London, London, United Kingdom 6. Leibniz Research Institute for Environmental Medicine, Düsseldorf, Germany Corresponding Author: Nino Künzli, MD, PhD, Deputy Director Swiss Tropical and Public Health Institute (SwissTPH); Department of Epidemiology and Public Health & Chair for Social and Preventive Medicine Medical School University Basel Mailing address: Swiss Tropical and Public Health Institute, P.O. Box, 4002 Basel, Switzerland; Tel: +41- (0) Fax: +41- (0) Nino.Kuenzli@unibas.ch Contact details of all authors: Benedicte Jacquemin, MD, PhD, researcher, Respiratory and Environmental epidemiology, CESP Centre for research in Epidemiology and Population health, U1018, Inserm, Villejuif, France and Université Paris Sud 11, UMRS 1018, Villejuif, France. Mailing address: 16, avenue Paul Vaillant Couturier, Villejuif Cedex, France Phone: +33 (0) Fax: +33 (0) benedicte.jacquemin@inserm.fr Tamara Schikowski MD, PhD, postdoctoral researcher. Swiss Tropical and Public Health Institute, Basel, Switzerland Mailing address: Swiss Tropical and Public Health Institute, P.O. Box, 4002 Basel, Switzerland, Tel: +41- (0) Fax: +41- (0) tamara.schikowski@unibas.ch 1

6 Anne Elie Carsin, statistician, Centre for Research in Environmental Epidemiology, Barcelona, Spain Consortium for Research on Epidemiology and Public Health (CIBERESP), Spain; Hospital del Mar Research Institute (IMIM), Barcelona, Spain. Mailing address: Parc de Recerca Biomèdica de Barcelona, Doctor Aiguader, Barcelona; Phone: Fax: Anna Hansell, MBB CHIR, PhD, Imperial College London, London, United Kingdom MRC-HPA Centre for Environment & Health Mailing address: Imperial College London, St Mary s Campus, W2 1PG, London, UK Phone: +44 (0) a.hansell@imperial.ac.uk Ursula Krämer, PhD, epidemiology, IUF - Leibniz Research Institute for Environmental Medicine, Düsseldorf, Germany Mailing address: Auf'm Hennekamp 50, Düsseldorf, Germany Phone: +49 (0) Fax: +49 (0) kraemeru@uni-duesseldorf.de Jordi Sunyer, MD, PhD, researcher, Centre for Research in Environmental Epidemiology, Barcelona, Spain Consortium for Research on Epidemiology and Public Health (CIBERESP), Spain; Universitat Pompeu Fabra, Barcelona, Spain; Hospital del Mar Research Institute (IMIM), Barcelona, Spain. Mailing address: Parc de Recerca Biomèdica de Barcelona, Doctor Aiguader, Barcelona, Spain Phone: Fax: jsunyer@creat.cat Nicole Probst-Hensch MD, PhD, researcher, Swiss Tropical and Public Health Institute Basel, Switzerland Mailing address: Swiss Tropical and Public Health Institute, P.O. Box, 4002 Basel, Switzerland Tel: +41- (0) Fax: +41- (0) Nicole.Probst@unibas.ch Francine Kauffmann, MD; researcher, Respiratory and Environmental epidemiology, CESP Centre for research in Epidemiology and Population health, U1018, Inserm, Villejuif, France and Université Paris Sud 11, UMRS 1018, Villejuif, France. Mailing address: 16, avenue Paul Vaillant Couturier, Villejuif Cedex, France Tel (33) Fax (33) francine.kauffmann@inserm.fr 2

7 Nino Künzli, MD, PhD, Deputy Director Swiss Tropical and Public Health Institute (SwissTPH); Department of Epidemiology and Public Health & Chair for Social and Preventive Medicine Medical School University Basel Mailing address: Swiss Tropical and Public Health Institute, P.O. Box, 4002 Basel, Switzerland; Tel: +41- (0) Fax: +41- (0)

8 Abstract The causes of adult-onset asthma are poorly established and the asthmogenic role of air pollution has been investigated primarily in children. This review assesses the current evidence of the association between air pollution and asthma incidence among subjects free of asthma at least until late childhood. Seven publications from five study populations fulfilled the inclusion criteria (1 case-control and 6 cohort studies). All but one used markers of local traffic-related air pollution to characterize long-term exposure. Those studies reported similar associations with traffic-related air pollution. However, protocols, definitions of asthma, and exposure assignment were rather heterogeneous, and three publications relied on the same study, thus, we abstain from meta-analytic summaries. Reported patterns of effect modification (e.g. by sex, atopy, or smoking) were inconsistent. Overall, the role of trafficrelated air pollution in adult-onset asthma is less conclusive than in childhood asthma. Larger studies with more consistent definitions of phenotypes and exposure assessment for local traffic-related pollutants (e.g. ultrafine particles) are needed. Pooling existing cohorts such as in the ongoing European ESCAPE and TRANSPHORM consortia are promising steps. There is, however, a need for large-scale mega-cohorts to investigate these effects in standardized ways, and to identify the most susceptible populations. Key words: Asthma incidence, air pollution, adults, traffic, ultrafine particulate matter 4

9 Introduction Asthma incidence has increased in the last decades, and although the epidemic may have reached a plateau in some Western countries, it remains a public health concern affecting between 5 and 10% of the general population (1). While genetic factors certainly play a role in asthmagenesis (2-3), the secular trends indicate high relevance of environmental factors. Investigating the contribution of environmental exposures to the onset of asthma is a challenge not only due to the complexities of the exposome (4) but as well in light of the phenotype as a range of obstructive pathologies may be grouped under the clinical phenotype of asthma (5). Asthma appears often in childhood, may stay inactive for long time and reappear or worsen later in life. However, the notion that asthma may always begin during childhood is not supported by more recent studies reporting new onset of asthma occurring during adulthood (6). Indeed in the ongoing debate about asthma possibly consisting of various phenotypes, adult onset asthma may be seen as a distinct phenotype and under the novel concept of categorizing disease entities by underlying mechanisms or endotypes separate consideration of adult-onset asthma in etiologic research is preferable (7). Genetic studies indicate as well differences between childhood and adult-onset asthma (3). However, apart from occupational asthma, which might be responsible for some 10-25% of adult asthma (8), the causes of asthma onset in adults have not been extensively investigated although the incidence rate of adult onset asthma (i.e. onset after age 16) is also not negligible (between 1 to 4 per persons per year) (9). A few studies have described other determinants such as female gender, lung function, atopy, obesity, and parental history of asthma (6, 10). To investigate the possible causal role of ambient air pollution in the development of adult-onset asthma, it is important to distinguish acute effects of air pollution from the longterm contribution of pollutants to the development of the underlying pathology. Given that 5

10 partly different mechanisms may be involved in triggering exacerbations among asthmatics versus promoting the development of hyperreactive airways, this distinction is important in planning and interpreting studies. As it is well known that exposure to ambient air pollution is associated with asthma exacerbations (11-14), measured e.g. as asthma attacks or exacerbations of symptoms, increases in medication, emergency visits, or hospitalization (15-17), the investigation of associations between long-term cumulative exposure to air pollutants and asthma onset need to rely on other outcomes than these events. Candidates are the prevalence or preferably the incidence of doctors diagnosed asthma or the course of chronic preclinical markers of the phenotype. The literature of the long term effect of air pollution on asthma onset is dominated by children studies (18-19). In children, proximity to traffic is associated with an increase in asthma prevalence and incidence (20-22), with substantial evidence for a causal asthomgenic role of the pollutants occurring in high concentrations along busy roads (19, 23-24). In contrast, in the 2010 Health Effects Institute report on traffic related air pollution, the expert panel concluded that the evidence between exposure to traffic-related pollution and adult-onset asthma was inadequate and insufficient as only one study was available (25). The biological mechanisms that could explain the association between air pollution and asthma incidence include chronic airway inflammation mediated by direct local irritation and by oxidative stress (26). Air pollution is also associated with lung function growth (27), levels (28-29) and decline (30). All these features are implicated in both asthma severity and etiology. The objective of this paper is to review current evidence of adult-onset asthma and air pollution. Methods Through PubMed search, we reviewed original articles in peer-reviewed journals published 6

11 until June The term searches were asthma and air pollution or traffic and incidence or onset, setting the limits to the articles in English and in adults. The search resulted in 576 papers, for which we screened the title, the abstract and the full article when necessary to identify the papers according to the criteria described below. In addition, we compared the list with recent reviews based on publications up to Nov 2009 on air pollution and respiratory health (18-19, 23-24) confirming that our search did not miss any eligible paper. We only included longitudinal or case-control studies that investigate the determinants of adult-onset asthma among those free of asthma at least until late childhood. Underreporting of childhood asthma at baseline survey of adults may overestimate adult-onset asthma among those with recurrent asthma in the follow-up investigations. Thus, information about age of onset is important to reliably define adult-onset asthma. Cross-sectional associations between air pollution and adult-onset asthma prevalence may be a proxy for the investigation of adult onset incidence if age of onset is known, if relapse and remission were constant, and if there is no self-selection of residential location after asthma onset (18). As these are rather unknown factors, we focus our review on incidence studies. However, we will discuss the use or limitations of prevalence in the discussion, providing a summary table of the three adult asthma prevalence publications (based on two studies) in the Supplement. A Canadian study in press at the time this review was submitted is not included in the review either although asthma incidence in young adults (N=395) who were free of asthma during childhood was considered as an outcome in the follow-up of this cohort of >3 000 children first examined in (31). Assignment of air pollution exposure was not clearly described and analyses were based on dichotomous air pollution exposure categories (high versus low). This limits the quantitative interpretation and comparison of both, the statistically significant findings seen for SO2 and the null findings reported for total suspended particles and PM3.3 in association with adult incident asthma. 7

12 Under the null hypothesis, incidence of adult onset asthma should be associated with the past long-term exposure to air pollutants. We, thus, included in this review all studies investigating the association between objectively measured markers of exposure to ambient air pollution and adult-onset asthma. Studies and results based on reported exposure to traffic were disregarded (32). Studies dealing with childhood asthma are not included (for reviews see (19, 23-24) ), neither were investigations on acute manifestations of adult asthma (such as wheezing, exacerbations or chronic bronchitis symptoms) in association with air pollution, ecological comparisons, nor studies dealing only with indoor air pollution. Studies of other chronic obstructive respiratory diseases such as COPD were also excluded. Results In total, seven publications from five study populations fulfilled the inclusion criteria. We present them in the order of publication date. The studies are summarized in table 1 and the main results presented in table 2. The earliest study identified to report an association between long term exposure to air pollution (namely ozone) and asthma incidence was a cohort of non-smoking California Seventh-Day Adventists assessing the long-term health effects of ambient air pollutants (AHSMOG) (33). Subjects were recruited in 1977, aged between 25 and 87 and followed up in 1987 and All of them were current non-smokers, while a 36% of males and 14% of females were ex-smokers. Air pollution estimates were derived using monthly interpolations from the three nearest fixed-site monitoring stations in California to the residential and working place zip code centroids. The monitoring network was not dense at these times, thus, while this reflects a personal assignment of ambient concentrations, it is a proxy of personal exposure to urban background pollution rather than a measure of local traffic-related exposure conditions. Monthly indices were averaged, cumulated and then averaged for the last 20 years 8

13 as they were required at baseline to have lived at least for the previous 10 years in the study area. In this analysis (33), they included subjects with available estimates of air pollution exposure who had lived at least 10 years in the study area before recruitment. Asthma incidence was defined with a positive answer to the question Has a doctor ever told you have asthma? in any of the two follow-ups but not at baseline. More than 80% of the new cases of asthma reported at some point of the follow-up also symptoms compatible with a diagnosis of asthma. In males, reported doctor diagnosed asthma was significantly associated with the 20- year mean 8-h average ambient O 3 concentration (RR 2.1 ( ) per 27ppb increase) (Table 2). No associations were found in women, nor with other measures of urban background pollution (PM 10, SO 4, NO 2 and SO 2 ). Previously in the same population, an association between total suspended sulfates and new cases of asthma was reported (34). They also found associations between both long term exposure to O 3 and total suspended particles (TSP) with several respiratory symptoms; however, these cannot be considered markers of incidence (35-36). The study by Modig and colleagues (25) was the first to investigate the contribution of traffic-related pollution to adult-onset asthma. In this study, traffic was characterized by measurements of outdoor NO 2 and traffic flow at the subjects residences. There were 203 clinically examined incident cases of asthma among the population of a Swedish town aged 20 to 60 years ( inhabitants). The cases were age- and sex-matched with 203 control subjects without asthma selected from the Swedish population registry. The asthma diagnosis was based on a standardized clinical examination. The subjects with asthma had a tendency to have higher residential exposure to traffic flows and home outdoor concentrations of NO 2 compared with the controls; however, the association did not reach statistical significance and was only borderline significant for the 67 cases with positive skin tests (OR 9

14 1.2 ( ) per 1 ug.m-3) (Table 2). Models included adjustment for smoking in this population with some 25% of current smokers and 50% never smokers. Modig et al. (37) conducted another study including participants from three Swedish cities included in the Respiratory Health in Northern Europe cohort (RHINE) (38). Subjects were recruited in 1990, aged between 18 and 45 years old, and followed-up in Exposure at each participant s home was estimated with dispersion models using winter half year concentrations of NO 2. They also used the distance between the participant s home address and the nearest major road. They defined subjects with asthma onset as those who answered no to asthma symptoms in the last 12 months and to asthma medication in the last 12 months at baseline and yes to any of those question plus yes to ever asthma at follow-up. In a stricter definition age of asthma onset was required to be after the first survey. There were 107 asthma-onset cases but only 55 incident cases based on the stricter definition. There was a positive association between both, asthma onset and more strictly defined asthma incidence, and the levels of NO 2 (OR 1.5 ( ) and 1.5 ( ) respectively per 10 ug.m- 3 increase) as well as living within 50m meters of a major road (OR 2.9 ( ) and 3.8 ( ) respectively) (Table 2). Results were adjusted for smoking in this population with close to 50% never smokers. In the European Respiratory Health Survey (ECRHS) (39-40) associations between NO 2 and asthma incidence were investigated in three interrelated publications (41-43). Subjects were recruited , at age 20 to 44 years, and followed up 8-12 years later. In both analyses more than 4000 adults from 17 cities from seven countries were included (Sweden (Gotebborg, Umeå and Uppsala), United Kingdom (Ipswich and Norwich), Germany (Erfurt), Belgium (Antwerp), France (Paris and Grenoble), Italy (Verona, Pavia and Torino) and Spain (Albacete, Barcelona, Huelva, Galdakao and Oviedo)). Subjects home addresses were geocoded and linked to outdoor NO 2 estimates, considered a marker of local traffic related 10

15 pollution. This information was obtained from the 1-km 2 background NO2 surface modeled in APMoSPHERE (Air Pollution Modeling for Support to Policy on Health and Environmental Risk in Europe) (44). In total, 183 incident cases were identified, with 95 confirming that the onset occurred between both surveys. Some 42-45% were never-smokers in these ECRHSbased analyses, and models adjusted for smoking status. In the first paper, asthma incidence was defined in the classic way (43) where cases were answering positively to the question: Have you ever had asthma? in ECRHS II among those subjects who answered no to this question in ECRHS I. The association between asthma incidence and NO 2 was stronger when restricting the analysis to those who reported an age of onset between both surveys (thus, with a coherent age of onset) but it reached statistical significance only in the total sample (OR 1.4 ( ) per 10 ug.m-3 increase) (Table 2). Associations in non-smokers were stronger. In an further analysis, the asthma definition was based on the asthma score previously proposed (45-46) as an alternative way of defining asthma (42). Although the score is based on symptoms otherwise not included in the review we consider it as this definition of asthma is based on a whole cluster of symptoms, used as a novel approach to define asthma phenotypes on a continuous scale rather than as a dichotomous entity (45-46). As shown, a high score is very specific for doctors diagnosed asthma. The analyses of Jacquemin et al. (42) used the number of reported respiratory symptoms out of five: wheeze and breathlessness, feeling of chest tightness, attack of shortness of breath at rest, attack of shortness of breath after exercise, and woken by attack of shortness of breath during the last 12 months (45-46). The score was derived in ECRHS II in subjects without asthma or asthma symptoms in ECRHS I. An incident case was one with a 0 score and no asthma ever at baseline but at least a score of 1 at follow up. Based on the score, 387 subjects had at least one symptom and 123 had at least two at follow-up. A significant association between traffic related air pollution and the 11

16 asthma score in those with no asthma and no symptoms at baseline was observed (Ratio of the mean score (RMS) 1.2 ( ) per 10 ug.m-3 increase) (Table 2). Associations with trafficrelated pollution were strongest for those moving from score 0 to a score!3 (RMS 2.57, 95% CI ), thus reporting a group of symptoms shown to be strongly predictive of doctor s diagnosed asthma (46). Significant interactions with smoking status was reported with stronger associations in non-smokers. The third analyses of ECRHS data investigated the role of genetic polymorphisms potentially modifying the association between NO 2 and asthma incidence and prevalence (41). Prevalence of current asthma was defined as a positive response to either of two questions in ECRHSII: attack of asthma during the last 12 months or current use of asthma medication. Asthma incidence was defined as reporting asthma (using the same definition as above) in ECRHSII excluding people who reported asthma or a history of asthma in ECRHS I. They evaluated polymorphisms in genes involved in oxidative stress pathways [gluthatione S-transferases M1 (GSTM1), T1 (GSTT1), and P1 (GSTP1) and NAD(P)H, quinine oxidoreductase (NQO1)], inflammatory response [tumor necrosis factor " (TNFA)], immunologic response [Toll-like receptor 4 (TLR4)], and airway reactivity [adrenergic receptor #2 (ADRB2)]. A significant interaction was found between NQO1 rs and NO2 for asthma prevalence and newonset asthma (p = 0.04) (Table 2). In the Swiss Cohort Study on Air Pollution and Lung and Heart Diseases in Adults (SAPALDIA) the association between the 11-year change in home outdoor traffic related PM 10 was significantly associated with asthma incidence among 2725 never smokers (hazard ratio 1.3 ( ) per 1 ug.m-3 change in traffic related PM10) (47) (Table 2). The air pollution estimates at both surveys were predicted and interpolated to the home s addresses using dispersion models that took into account source specific emission and meteorological data. The population based sample of adults (age 18-60) was recruited in 1991 and followed-up in 12

17 2002. Asthma incidence was defined as positive answers to both Have you ever had asthma? and was this confirmed by a doctor? at the follow-up in those without asthma or COPD at baseline. The reported age of onset had to be after the first survey, resulting in 41 new cases of asthma. Discussion We found seven publications based on data from five adult asthma studies (Table 1). The prospective studies showed rather consistent positive associations between metrics of long term exposure to ambient air pollution and asthma incidence (Table 2). Six of them (from four studies) found associations with markers of traffic-related local air pollution, namely NO (25, 37, 42-43) 2 and modeled traffic-related particulate matter (47). One study compared traffic intensity measures and NO 2 finding stronger associations for the former (37). Due to the heterogeneity in the definitions of asthma, the differences in the exposure assessment (discussed later), and the lack of independence of estimates (with three analyses based on ECHRS), we abstain from a meta-analysis of the results summarized in table 2. A very recent systematic review considering asthma incidence concluded that air pollution may play a causal role in the development of asthma (19) however several differences to our review need to be taken into account. The aim of our review was to focus only on adult-onset asthma whereas Anderson et al (19) summarized the literature across all ages. The pathophysiologic mechanisms leading to the chronic phenotype of asthma are complex and many remain unknown. Childhood and adult onset asthma may be partly different phenotypes (3, 48-49). Under this concept one may conjecture differences in the causal patterns and underlying mechanisms of the various disease entities or endotypes labeled as asthma (7), thus, research may need to assess the 13

18 causes of childhood and adult onset asthma separately. The rather strong evidence for a causal role of air pollution in the development of childhood asthma (19) may not be generalized to adults without studies conducted in this age group. Our review focuses strictly on studies where incidence was defined as a chronic condition (defined e.g. as reported doctors diagnosed asthma ) or in one case based on new onset of a cluster of asthma symptoms among those with no asthma and no symptoms at baseline (42). We excluded studies (or results reported in the studies included in our review) where asthma was defined through the expression of exacerbations (e.g. period prevalence of single symptoms such as wheezing). As also emphasized by Anderson et al, there is strong evidence that exacerbations can be caused by acute peaks of air pollution exposure among many factors. This evidence of acute effects of air pollution on asthma has implications for the approach and ability to investigate the long-term contribution of air pollution to the development of asthma. Assuming air pollution was only a trigger of exacerbations, e.g. triggering wheezing episodes, then the period prevalence of wheezing would be associated with the average (e.g. annual mean) concentrations of ambient air pollution, since the annual or any long-term mean ultimately depends on the daily levels of pollution. In other words, asthmatics living at a more polluted residence will have a higher probability to report a wheezing history (e.g. during the last 12 months ) than asthmatics living at a cleaner site even if air pollution had only acute effects on wheezing. It is for that reason that symptom frequencies or any other expressions of acute asthma-related problems such as attacks, increased use of inhalers, or hospitalizations can hardly be used to unambiguously investigate the role of air pollution in the development of asthma. Accordingly, we have also not included recent analyses of the large Danish Diet, Cancer, and Health (DCH) cohort (50). In that innovative analyses, long term average concentrations of modeled home outdoor NO 2 were associated with an increase in first-ever 14

19 asthma related hospitalization in the elderly. Asthma incidence among the adult participants (age 50 to 65 at baseline, in ), was defined as a first-ever hospitalization due to asthma occurring during the 10.2 years of median follow-up time (until 2006). There were 821 subjects (1.5%) experiencing such a first hospitalization. While the exposure assessment was rather unique taking into account 35 years, those long-term mean NO2 values were highly correlated with the means across shorter periods, including the last year prior to hospital admission. As a consequence, models could not include both, acute and long-term exposure terms to possibly disentangle the contributions of different time windows of exposure. It is thus impossible to clarify whether the observed associations reflect an estimate of the acute effects versus a long-term increase of the pool of asthmatics due to long-term exposure to pollution. An increase of this pool would likely result in increased asthma-related hospitalizations, irrespective of current pollution (51). Moreover, defining asthma incidence by the occurrence of hospital admissions captures only the most severe subset of adult-onset asthma as not all asthmatics experience hospitalizations. We have also not included studies reporting higher symptom rates in population based samples (thus, including non-asthmatics and asthmatics) such as the SAPALDIA analyses (52-53) included in the review by Anderson et al (19). Although symptoms are a key feature leading to doctor s diagnosed asthma, a symptom period alone is not sufficient to define asthma, and none of these symptoms are per se a good proxy for asthma prevalence or incidence. Moreover, the observed associations with pollution may again reflect increased period prevalence due to the sum of all acute effects of air pollution experienced during the past 12 months. We did instead include the ECRHS analysis based on the score. Although this is symptom-based as well, it is not the association of pollution with the period prevalence of a single symptom but indeed with the change from a state without asthma or symptoms into a 15

20 state characterized by a cluster of symptoms shown to have a very high positive predictive value for doctors diagnosed asthma (45). This is not the case for any single symptom and the fact that the strongest associations were observed for the change from score 0 to!3 symptoms supports the interpretation as asthma incidence. Exposure assessment methods were heterogeneous across studies (Table 1). Accurate and relevant exposure assessment is fundamental to assessing relationships of asthma onset with air pollution. As seen in the literature on childhood asthma, studies that did not describe local traffic-related pollution with its known substantial within-community gradients consistently fail to see any association instead with urban background pollutants (18). The evidence of a causal role of air pollution in the development of childhood asthma is based entirely on studies assigning some markers of local traffic-related pollution (19, 23-24). Those pollutants (e.g. ultrafine particles, CO or NO) show substantial spatial gradients over short distances, within communities with a steep drop-off in pollutants to background levels within m distance from road (23). All the adult studies but AHSMOG (33) attempted to characterize the local home outdoor exposure conditions and specify local traffic-related pollution, using markers known to be related to those types of pollutants with high spatial within-city communities such as locally modeled NO2 (or NOx) (25, 37, 42-43), traffic-related particulate matter estimated from emission-based dispersion models (47), or distance to busy roads (25, 37). While all seven publications estimated air pollution concentrations at the individual level, i.e. at least the residential addresses, McDonnell (33) did so through interpolation to the closest monitors, which at that time were rather sparse. Thus, the study did certainly not capture local trafficrelated pollution, and, if the latter matters it is not surprising that the study finds no association with any marker of combustion related particulate pollution. In ECRHS (41-43) they used NO 2 estimates derived from dispersion models on a quite large 1x1km grid. While the 16

21 local contrasts in exposure are not well captured on a 1x1km grid either, local contrasts remained rather large, possibly due to the ECRHS restriction to only include cities with at least inhabitants rather than smaller towns. As shown in Anderson et al. (19) in children studies, associations between pollution and asthma were positive only in studies using markers of the local traffic-related pollution while urban background pollution was not associated with asthma onset nor prevalence. This discrepancy is particularly interesting in light of the fact that, for other respiratory (and cardiovascular) effects of ambient air pollution, such clear differences between associations based on urban background versus local traffic-related pollutants did not emerge. For example, in the SAPALDIA study, symptoms correlated both with the urban background level of pollution (53) and the more sophisticated locally assigned metrics (52). This may indicate that the specific toxicological properties of these pollutants relate to partly different mechanisms relevant in asthma development. Of interest is the distinction of prevalence and incidence studies. As conceptualized by Anderson et al. (18),asthma prevalence studies may, in theory, lead to the same conclusion as incidence studies if remission and relapse do not occur or are constant, thus, prevalence studies could support the assessment of evidence. Unfortunately, in the case of children studies, the vast majority of prevalence studies lack any information of traffic-related local air pollution, thus, the null findings for the prevalence studies (18) cannot be considered inconsistent with the positive findings on incidence studies which were mostly based on traffic-related pollution. However, in case of adult studies, the situation is different. There are no adult-onset asthma prevalence studies published except the three recent prevalence based publications conducted in Sweden by Lindgren et al. based on two different populations (54-56). In a study including 9319 adults from 18 to 77 years old from Malmö and surrounding areas, Lindgren at al. (55) evaluated the association between traffic related air pollution and asthma 17

22 prevalence (among other respiratory diseases, such as COPD or asthma symptoms). A positive answer to Have you been diagnosed by a doctor as having asthma? defined asthma prevalence. They did not have information on age of asthma onset. Objective markers of air pollution exposure were traffic intensity within 100m of the residence and NO x individually assigned from dispersion models on a grid of 250 m 2. The study reported that living within 100 meters of a road with heavy traffic (>10 cars/min) was associated with a higher prevalence of asthma diagnosis (OR 1.4 ( ). In a second publication in the same study, Lindgren et al. (56) have shown that living close to traffic was associated with current allergic asthma, i.e. asthma triggered by allergic factors such as pollen of furred animal, (OR 1.3 ( )) but not with non-allergic asthma. No association was found with home outdoor modeled NOx in any of the two analyses. Indeed, the analyses provided numerous estimates on various subgroups and measures of exposure with rather inconsistent patterns, and in the subgroup of suburban dwellers, NOx was inversely related to asthma prevalence. In another population, in a two steps case-control study, Lindgren at al. (54) assessed the association between traffic exposure and asthma prevalence. In the first step, adults (age 18-65) from Scania, for whom they had home addresses were included. In a second phase they conducted a nested case-control study of adults including 2856 participants who were asked for the work address and time spent in traffic. Complete weighted traffic exposure information was available for 1488 subjects. Exposure estimates were derived from NO x dispersion models, both at home and at work addresses. They also linked the addresses to the national road database in order to integrate exposure to traffic during commute in a model of total NOx exposure, reflecting the time-weighted average for home, work, and commuterelated concentrations. Asthma prevalence was defined as a positive answer to Do you have asthma? They also found that living within 50m of a road with high traffic was associated 18

23 with higher asthma prevalence (OR 1.8 ( )). However no association was found with any of the NO 2 based analyses (home, total, work or commute). Thus, in contrast to the incidence studies, these three publications on prevalence of Lindgren et al. provide conflicting results with suggestive associations seen for proximity to busy roads but null findings for local estimates of NO 2. Disease-based moving patterns and environmental preferences in the selection of residential location may affect prevalence studies more strongly than prospective studies but the findings are inconsistent and not easily explained. Mechanisms To elucidate the speculation that local traffic-related air pollutants (e.g. ultrafine particles) affect asthma development differently than e.g. larger particles one would need to understand the web of mechanisms causing asthma in general and the mediating pathways of air pollution. The most accepted mechanisms of lung damage caused by particulate matter are inflammation (mainly due to oxidative and nitrosative stress) and reduction of the defense capacity increasing susceptibility to infections (57). In children studies, it has been suggested that the mechanisms linked to asthma incidence include allergic sensitization. In particular diesel exhaust may act as an adjuvant affecting the immune response to environmental allergens. Nel emphasized the distinct properties and contributions of the ultrafine fraction of particles i.e. those with high spatial differences within communities and the larger PM fractions (58). The immunologic effects of particles may depend on the alveolar translocation which in turn heavily depends on particle size. However the importance of allergy is far less important in adult onset asthma, thus, the relevance of these immunologic pathways is less clear in adults (59) and interactions between pollution and markers of allergies were inconsistent across the studies reviewed. 19

24 Several studies used NO 2 as the marker of air pollution. This raises the question whether NO 2 per se could be causally implied. The toxic effect of NO 2 on the human respiratory tract at ambient concentrations is less important than for other gases (e.g. ozone) or particles. At an experimental level, as a free radical, NO 2 has the capacity of depleting the antioxidants of the lung tissue, causing subsequent injury and inflammation (60). Using animal or in vitro models, NO 2 produces eosinophilic inflammation, enhances epithelial damage, reduces mucin expression and increases baseline smooth muscle tone (60-63). Repeated exposure to high doses of NO 2 is associated with increased breath frequency and decreased lung distensibility and gas exchange (64). It has also been explained that NO 2 decreases bactericidal activity and alveolar macrophage activity (63, 65-66). Whether these experimental findings can be extrapolated to human environmental exposure conditions is not known. Heterogeneities and susceptible subgroups Asthma definitions, thus, possibly phenotypes, differed across studies although several used the same core questionnaires (namely ECRHS (42-43), SAPALDIA (47), and RHINE (37) ). All the studies defined asthma onset, excluding subjects with self-reported asthma and/or asthma like symptoms and/or doctor diagnosed asthma at baseline and designated as cases subjects reporting asthma at follow-up. All but one (Modig et al. (25) who used clinical examinations), used the answers from a questionnaire. However, studies differed in the strict definition of a new case ; not all required age of onset to be prior to the baseline survey meaning it may include cases reflecting remissions of childhood asthma rather than a truly new-asthmatics, thus, not appropriate for the investigation of causes of adult onset asthma. As reported in ECRHS (42-43), SAPALDIA (47), and RHINE (37), at least some 50% of all new cases may be such remissions of previously not reported childhood asthma. 20

25 Another issue to take into account in the heterogeneity between the populations from the five studies included in this review. In ECRHS (42-43) and RHINE (37) participants were younger than in the other studies. In SAPALDIA (47) the association between asthma incidence and air pollution was only observed in never-smokers and in the ASHMOG (33) only in nonsmokers while all other studies adjusted for smoking status. The first Modig paper (25), was a case-control whereas all the other population based. Susceptibility factors, in particular related to sex (hormonal factors), atopy, physical activity or diet (as modifiers of oxidant response), co-morbidities in particular COPD or life-styles such as smoking were not consistently addressed. Castro-Giner et al. (41) is the only study so far addressing gene-air pollution interactions. The findings strongly support the need to take biological susceptibilities into account. Three studies presented in this review addressed susceptibilities, the effect of air pollution on asthma seemed stronger in atopics in two studies (37, 47), while in non-atopics in the third (43) but none has enough power to show a significant interaction. Results concerning gender differences were also inconsistent across the studies, even if the majority seemed to find a higher association in males. In the ASHMOG study (33) associations were only seen in males. In RHINE (37) and SAPALDIA (47) the association was slightly higher in males while in ECRHS it was the same in both gender when using the asthma score (42) but was slightly higher in females when using the classical definition (43). However, statistical power was insufficient to evaluate these differences. In SAPALDIA (47), air pollution was more strongly associated with young atopic asthma in males. In SAPALDIA, positive findings were only observed in non-smokers and in ECHRS analyses, signals were also larger in non-smokers whereas the other studies give no clear information about interactions of pollution with smoking status. We agree with Anderson et al. (19) that none of the studies can clearly answer the question whether removal of air pollution exposure would result in fewer asthma cases or just 21

26 postponement of onset of asthma. This limitation is indeed true for most if not all established risk factors of chronic diseases. For example, smokers developing COPD might have done so also had they never smoked, but possibly at a later stage in life or in a less severe form. What is relevant from a public health perspective is that science-based preventive action e.g. through stringent implementation of clean air policies may at least prolong the disease free state. CONCLUSION AND FUTURE RESEARCH NEEDS The currently available studies on adult onset asthma are not sufficient to come to firm conclusions about the causal role of ambient air pollution. Based on the prospective studies, results would be much in line with a causal role of traffic-related air pollution in the development of adult-onset asthma consistent with that seen for asthma in children. However, the inconsistency in the two prevalence studies between the significant results for traffic density (proximity) but null findings for locally modeled NO 2 cannot be dismissed. Those NO 2 models may reflect similar traffic-related pollutants as those used in the adult incidence studies. It is worth noting that the three conflicting publications are all based on the same geographic area in Southern Sweden. However, if one gives stronger weight to the prospective incidence studies - less prone to self-selection bias evidence for a causal role of traffic-related pollutants is stronger. Many questions remain unanswered and should be tackled in future studies in adults. There is in particular a need to characterize more rigorously and consistently the exposure to local traffic-related pollution, including traffic-related exposure during commute and at work. Lindgren et al. (54) made innovative attempts to do so for NO 2. Given the null findings it is though difficult to interpret the lack of any difference among home-based versus home, work and commute based results (54). Better characterization of the ultrafine fraction of vehicle exhaust emissions is needed to better understand exposures to different particle types. This 22

27 includes the need to tackle ultrafine particle exposure during time spent in traffic or outdoors as this may contribute substantially to total exposure to traffic-related toxicants and be relatively more important than those exposures experienced at home (67). More rigorous and consistent definitions of the adult onset asthma phenotype should be part of further investigations as the role of traffic-related air pollution may differ across phenotypes. This includes the need for a coherent distinction between adult-onset asthma and remission of childhood asthma later in life. As discussed, adults may forget the reporting of childhood asthma until the disease becomes again active. Consistent consideration of possible susceptibility factors such as gender or atopy is also needed to understand those at highest risk, which is of both clinical as well as public health relevance. A major challenge to address all the open questions relates to the need of having sufficiently large prospective cohort studies. None of the published studies have the power to investigate interactions as the number of new cases of asthma was rather low. In the future, large mega-cohorts will provide opportunities to address these questions. Meanwhile a promising approach lays in the collaborative research combining comparable existing cohort studies. A promising initiative in this direction comes from the European ESCAPE and the related extension of the TRANSPHORM projects ( The role of traffic-related air pollution in the development of chronic respiratory diseases in adults will be investigated across six existing cohort studies, including SAPALDIA (47), ECRHS (43), SALIA (68), EGEA (69-70), E3N (71), and the U.K. National Survey of Health and Development (1946 birth cohort) (72). Acknowledgements: 23