You are here

  • Home
  • Archive
  • Volume 70, Issue 5
  • Acute nasal pro-inflammatory response to air pollution depends on characteristics other than particle mass concentration or oxidative potential: the RAPTES project

Article Text

Article menu
Environment
Original article
Acute nasal pro-inflammatory response to air pollution depends on characteristics other than particle mass concentration or oxidative potential: the RAPTES project
  1. Maaike Steenhof1,
  2. Ian S Mudway2,
  3. Ilse Gosens3,
  4. Gerard Hoek1,
  5. Krystal J Godri2,4,
  6. Frank J Kelly2,
  7. Roy M Harrison4,5,
  8. Raymond H H Pieters1,
  9. Flemming R Cassee1,3,
  10. Erik Lebret1,3,
  11. Bert A Brunekreef1,6,
  12. Maciej Strak1,3,
  13. Nicole A H Janssen3
  1. 1Division of Toxicology, Institute for Risk Assessment Sciences (IRAS), Utrecht University, Utrecht, The Netherlands
  2. 2MRC-HPA Centre for Environment and Health, King's College London, London, UK
  3. 3Centre for Environmental Health, National Institute for Public Health and the Environment (RIVM), Bilthoven, The Netherlands
  4. 4Division of Environmental Health & Risk Management, University of Birmingham, Birmingham, UK
  5. 5Department of Environmental Sciences, Center of Excellence in Environmental Studies, King Abdulaziz University, Jeddah, Saudi Arabia
  6. 6Julius Center for Health Sciences and Primary Care, University Medical Center Utrecht, Utrecht, The Netherlands
  1. Correspondence to Bert Brunekreef, Division of Environmental Epidemiology, Institute for Risk Assessment Sciences (IRAS), Utrecht University, P.O. Box 80178, 3508 TD, Utrecht, The Netherlands; b.brunekreef{at}uu.nl

Abstract

Objectives To investigate which air pollution characteristics are associated with biomarkers for acute nasal airway inflammation in healthy subjects. We hypothesised that associations would be strongest for oxidative potential (OP) of particles.

Methods 31 volunteers were exposed to ambient air pollution at five sites in The Netherlands: two traffic sites, an underground train station, a farm and an urban background site. Each subject visited at least three sites between March and October 2009 and was exposed for 5 h per visit including exercise for 20 min every hour (h). Air pollution measurements during this 5-h-period included particulate matter (PM) mass concentration, elemental composition, elemental and organic carbon (OC), particle number concentration, OP, endotoxins, O3 and NO2. Pro-inflammatory biomarkers were measured before, 2 and 18 h postexposure, including cytokine IL-6 and IL-8, protein and lactoferrin in nasal lavage (NAL) as well as IL-6 in blood. One- and two-pollutant mixed models were used to analyse associations between exposure and changes in biomarkers.

Results In two-pollutant models, cytokines in NAL were positively associated with OC, endotoxin and NO2; protein was associated with NO2; and lactoferrin was associated with all PM characteristics that were high at the underground site. In blood, associations with OC and endotoxin were negative.

Conclusions We observed no consistent effects in two-pollutant models for PM mass concentration and OP. Instead, we found consistent associations with nasal inflammatory markers for other PM characteristics, specifically OC, endotoxin and NO2.

  • Particulate Matter
  • Oxidative Potential
  • Inflammation
  • Nasal

https://doi.org/10.1136/oemed-2012-100993

Statistics from Altmetric.com

    Request Permissions

    If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

    What this paper adds

    • It is not well known which air pollution characteristics are responsible for adverse health effects.

    • This paper shows that acute air pollution exposure is associated with increased expression of biomarkers of nasal inflammation in healthy subjects.

    • In two-pollutant models we found consistent associations for organic carbon, endotoxin and NO2, but not for particulate matter (PM) mass concentration or oxidative potential of particles.

    • Our results suggest that air quality management should focus on specific air pollution compounds rather than on PM mass concentration which is currently used for air quality legislation.

    Introduction

    It is not well known which particulate matter (PM) characteristics are responsible for adverse health effects.1 ,2 Numerous characteristics, like particle size, number concentration, transition metals, organics, sulfates and nitrates, and biological components such as endotoxins have been proposed. Thus far, the available in vitro, in vivo and epidemiological data provide insufficient information to enable further quantification on the importance of individual PM characteristics.1 ,3

    Currently, attention is being given to the capacity of particles to exert oxidative stress as it is thought to be an important mechanism underlying PM-induced adverse health effects.4 Oxidative stress results when the balance between the generation of reactive oxygen species, or free radicals, overrides the cells’ antioxidant defences. High levels of oxidative stress induce inflammatory responses via a cascade of events including activation of various transcription factors and stimulation of cytokine production.4 ,5 Since the oxidative potential (OP) of particles reflects several of the physical and chemical properties that contribute to PM toxicity, it may be a biologically meaningful, integrative measure to predict PM-related health effects.6 ,7

    Being the entry port for inhaled PM, the nasal cavity is a primary target for effects from ambient particles. In the nasopharyngeal cavity, deposition rates are the highest for the coarse size fraction (∼70%) and those particles in the ultrafine size fraction that are about 1 nm in diameter (∼90%).8 Deposited particles can trigger nasal cells to release pro-inflammatory mediators (eg, cytokines, reactive oxygen species) or cause cell damage to the epithelial cells lining the nasal airways. Other responses characteristically for inflammation include local recruitment and activation of cells, and increased vascular permeability. Several studies have been published that examined nasal responses in relation to air pollution exposure.9–11 However, these studies were not designed to quantify the health effects of individual PM components or PM emissions from different sources.

    This study is part of the RAPTES project: Risk of Airborne Particles—a hybrid Toxicological and Epidemiological Study. The aim of RAPTES was to assess the independent contribution of specific PM characteristics to acute cardiovascular and respiratory health effects. Next to in vitro and in vivo experiments,12 we performed an extensive volunteer study of healthy subjects semiexperimentally exposed to ambient PM at real-world locations with different PM characteristics.13–15

    Here, we measured interleukin (IL)-6 and IL-8, as well as total protein and lactoferrin concentrations in the fluid obtained from nasal lavages (NAL). Cytokines IL-6 and IL-8 were selected because they are key players in the process of inflammation and known to increase in bronchoalveolar lavage fluids of healthy volunteers exposed to diesel exhaust.16 ,17 Total protein is a traditional marker of airway injury, and if measured in NAL it can be indicative for cell damage as well as increased epithelial and vascular permeability.18 In contrast to the other select markers, lactoferrin is not evidently linked to nasal inflammation or frequently measured in air pollution exposure studies. Lactoferrin is a multifunctional iron-binding protein, known to exhibit both pro-and anti-inflammatory and oxidative stress-related activities. It is primarily synthesised and secreted by glandular cells present in (nasal) epithelium, but also found in the secondary granules of neutrophils.19 This marker was selected because we wanted to investigate the relationship between lactoferrin concentrations and particle properties like OP and transition metal content (iron in particular).

    The aim of this study was to investigate the independent contribution of specific air pollution characteristics to changes in nasal biomarkers. We hypothesised that short-term PM exposure induces increased expression of nasal airway inflammation markers in healthy volunteers and that this effect would be most strongly associated with PM OP.

    Methods

    Study design

    The RAPTES study design was explained previously.14 In short, we used a semiexperimental design exposing healthy volunteers to ambient air pollution at five different sites in The Netherlands: a continuous traffic site, a stop-and-go traffic site, an underground train station, a farm and an urban background site. The rationale for selecting different sites was to create high contrast and low correlations among different air pollutants.13 Site visits were performed on 30 weekdays from March to November 2009. Volunteers were healthy, young, non-smoking students living at the campus of Utrecht University. Subjects participated in 3–7 visits scheduled at least 14 days apart for each individual. Exposure started around 09:00 and lasted for 5 h, during which detailed characterisation of PM air pollution was performed on-site. Subjects performed moderate exercise (minute ventilation 20 l/min/m2) on a bicycle ergometer for 20 min every hour. Before and at 2 and 18 h after exposure, we collected NAL and blood samples to measure various biomarkers of (nasal) inflammation. To minimise exposure during transport of participants to the sampling locations, we equipped a minibus with a custom-made cabin air filter and measured particle number concentration (PNC) with a portable condensation particle counter (TSI CPC3007) during each commute. The study was approved by the Medical Ethical Committee of Utrecht University and each subject gave written informed consent.

    Exposure assessment

    We previously described in detail the measurements during the on-site exposures.13 In summary, we used Harvard Impactors to measure PM10 and PM2.5 mass concentration. The mass concentration of PM2.5–10 was calculated as the difference between PM10 and PM2.5. PM10 and PM2.5 samples were used to determined absorbance (as a measure for soot) and endotoxin concentration (PM10 only). PNC was measured using a condensation particle counter. Gaseous pollutants concentrations (O3 and NO2) were measured using real-time monitors. Coarse (C) and fine (F) samples collected with a high volume sampler were measured for the concentrations of elemental carbon (EC) and organic carbon (OC), trace metals (both water-soluble and ‘total’ acid-extracted), and secondary inorganics (NO3− and SO42−). We report absorbance, NO3−, SO42− in the fine fraction, while the individual PM fractions of trace metals were aggregated. Particle intrinsic OP was measured in vitro by depletion of antioxidants ascorbate (AA) and glutathione (GSH) in synthetic human respiratory tract lining fluid. The sum of both metrics is presented as OPTOTAL.20 OP was analysed for the coarse, fine and quasi ultrafine size fractions of PM sampled with a micro-orifice impactor. In this paper, we report OPTOTAL aggregated for the size fractions. Results of the individual metrics (OPAA and OPGSH) per size fraction are shown in the online supplementary material.

    Inflammation markers in NAL

    Subjects were asked to sit, move the chin towards the chest, close one nostril with their hand and mouth breathe. Then 5 ml sterile phosphate buffered saline (37°C) was instilled in the other nostril via a nasal olive connected to a 30 ml syringe. The fluid was held for 5 s, and expelled through a funnel with polyamide gauze into a centrifuge tube. Next, the procedure was repeated for the other nostril using the same tube. The first two pre-exposure lavages were discarded to minimise possible wash-out effects between pre-exposure and postexposure (PE) lavages. The third lavage, collected after 30 s per nostril, was collected in a separate tube and regarded as the baseline sample. PE lavages were collected at 2 and 18 h after exposure in the same manner (one lavage per nostril). All NAL samples were successfully obtained and put on ice immediately after collection. Samples were centrifuged for 10 min (250 g) at 4°C. The supernatant was aliquotted into a series of nine vials that were stored at −80°C until further analysis after the fieldwork ended.

    Pre-exposure, the recovery volume of NAL fluid was 8.1 ml (1.6 ml) expressed as median (IQR). PE recovery volumes were 7.5 ml (1.4 ml) and 7.5 ml (1.7 ml) for the 2 and 18 h time-point, respectively. Commercial ELISA kits were used to measure IL-6 and IL-8 (R&D Systems, Abingdon, UK), and lactoferrin (Hycult Biotech, Uden, The Netherlands). Total protein was determined with the BCA protein assay (Pierce, Rockford, Illinois, USA). None of the samples were below detection limit (0.04 pg/ml, 3.5 pg/ml, 0.4 ng/ml and 20 µg/ml, respectively). In some cases, not all markers could be analysed due to low sample volume.

    Inflammation markers in blood

    The blood sampling procedure is described in detail by Strak et al.15 Blood collection was successful in 503 out of 510 donations. IL-6 was measured in serum (R&D Systems, Abingdon, UK).

    Data analysis

    We analysed associations between air pollutants and biomarkers using mixed linear regression to account for correlation between measurements on the same subject across different visits, using a random intercept model with compound symmetry as covariance structure. The dependent variables were the changes in biomarkers between PE (2 and 18 h) and pre-exposure measurements in our study population. Biomarker levels were log-transformed to normalise distributions when needed. This was the case for all biomarkers except for protein in NAL. The independent variables were the 5 h average on-site concentrations of air pollutants. As potential confounders, we included on-site temperature and relative humidity, and season. Transport to and from the sampling sites did not affect associations with the on-site exposures and was therefore not included as confounder.

    Details of our analysis strategy have been published by Strak et al.14 We specified both one- and two-pollutant models to assess the individual effect of each pollutant as well as the stability of the pollutants’ effects. Since the underground site had substantially higher concentrations of most pollutants, we also analysed the data separately for the outdoor sites (outdoor dataset; excluding the underground site). Endotoxin was much higher at the farm, and we therefore also performed a separate analysis excluding the farm (without farm dataset; excluding the farm site).

    Since we defined a large number of models, we considered an association consistent if the p value in the one-pollutant model was smaller than 0.1 and remained so after adjusting for all other co-pollutants in two-pollutant models, or alternatively, if the p value in the two-pollutant models remained smaller than 0.1 for those models in which the co-pollutant was not highly correlated with the pollutant of interest (Spearman's R>0.7).

    The impact of influential observations was assessed using the Cook's D statistic as described in the online supplementary material. Associations with lactoferrin in NAL were highly influenced by one single observation. For this biomarker, we therefore present results from the Cook's D analysis (excluding 1% of the most influential observations) instead of the analysis including all observations. Associations with other end points were not as highly influenced.

    Sensitivity analysis was performed by excluding observations of participants who (a) reported nasal allergies, (b) were former smokers and (c) reported symptoms of upper airway inflammation in the 24 h period before the start of the sampling day. The observed associations were similar to those observed in the complete dataset (data shown in the online supplementary repository, RAPTES 2013).

    Effect estimates and their CIs were presented as percentage increases over our study population mean of the baseline values. We expressed these percentage increases per changes in IQR for the outdoor locations and, for endotoxin, for all locations other than farm to allow direct comparison of effect estimates across the various analyses.

    All data analyses were performed using SAS V.9.2 (SAS Institute, Cary, North Carolina, USA).

    Results

    In total 170 observations were obtained on 30 sampling days: 9 days at the underground train station site (n=45 observations), six at the stop-and-go traffic site (n=37), and five at each of the farm, continuous traffic and urban background sites (n=28, n=31 and n=29 respectively). The study population characteristics are shown in table 1. Baseline levels of NAL biomarkers varied significantly between subjects (p<0.001) except for lactoferrin.

    View this table:
    Table 1

    Descriptive characteristics of the study population at the baseline (t=0)

    Exposure assessment

    Air pollution data are shown in online supplementary table S1. At the underground location, concentrations of nearly all PM characteristics were substantially higher than at the other sites. In contrast, endotoxin levels were substantially higher at the farm. Spearman correlations between the measured pollutants can be found in the online supplementary repository (RAPTES 2013) and in Strak et al.14 Briefly, in the complete dataset, PM10, PM2.5, absorbance, EC, OC, trace metals and OP were highly correlated (Spearman's R>0.7). Correlations between endotoxin and all other pollutants were low except for OC(C) (Spearman's R 0.59). Those high correlations decreased considerably after excluding the underground train station data.14 Correlations between components in the without farm dataset were mainly similar to those observed in the dataset including all sites except for endotoxin and OC(C) (Spearman's R 0.49).

    View this table:
    Table 2

    Associations between air pollution exposure and percentage changes (post–pre) in NAL IL-6 2 h after exposure in the complete dataset including all sites

    Associations with NAL IL-6 and IL-8

    Results of one-pollutant models are shown in the online supplementary repository (RAPTES 2013). Both IL-6 and IL-8 in NAL were significantly positively associated with several pollutants. Across the two cytokines and time-points (2 and 18 h PE) we found the strongest associations with endotoxin, OC(C) and NO3− in both the complete dataset and after excluding the underground. All of these associations became non-significant after excluding data from the farm, after which a significant positive association with several traffic-related components (especially PNC and NO2) became apparent. We also observed several significant negative associations, but these effects were less consistent across the two cytokines, time-points or datasets.

    The results of two-pollutant models measured 2 h PE are shown in this paper and the online supplementary material, and the results of two-pollutant models measured 18 h PE are shown in the online supplementary repository (RAPTES 2013). Two hours PE, in the complete dataset, we found consistent positive associations among NAL IL-6 and IL-8, with NO3− and endotoxin (table 3 and see online supplementary tables S2 and S3). Associations for OC(C) only lost significance after adjusting for NO3−. In addition, we observed a consistent significant negative association with water-soluble V for IL-6, which was not observed for IL-8.

    View this table:
    Table 3

    Associations between air pollution exposure and percentage changes (post–pre) in NAL protein 2 h after exposure in the complete dataset including all sites.

    In the outdoor dataset, the associations for OC(C) and endotoxin with both IL-6 and IL-8 were consistent, but the effect of NO3− disappeared after adjusting for OC(C) (see online supplementary tables S4 and S5). In the without farm dataset, the significant positive associations for NO2 with IL-6 and IL-8 were consistent, while the effect of PNC was only observed with IL-6 and lost significance after adjustment for NO2. In addition, we found a fairly consistent significantly negative association for O3. However, as for PNC, these effects were only observed with IL-6 and became non-significant after adjustment for NO2 or PNC (see online supplementary tables S6 and S7).

    Eighteen hours PE, associations remained for OC(C), endotoxin and NO2, albeit weaker and less consistent compared with the earlier time-point. There were no (fairly) consistent significant associations with any of the other pollutants measured, including NO3− and PNC (see online supplementary repository, RAPTES 2013).

    Associations with NAL protein

    In one-pollutant models (see online supplementary repository, RAPTES 2013), in both the complete dataset and the without farm dataset, we found significant positive associations with NO2 (2 h PE), and significant negative associations with water-soluble Fe (2 and 18 h PE). At 2 h PE, the association with water-soluble Fe was only significant after excluding 1% of the most influential observations (Cook's D analysis, effect estimate increased by 50% and p values changed from >0.1 to <0.05). In the outdoor dataset we found significant positive associations with several other pollutants 2 h PE, but only NO2 and absorbance were significant at the p<0.05 level.

    In two-pollutant models, the positive associations with NO2 2 h PE were consistently significant in all datasets (table 3 for all sites; see online supplementary tables S8 and S9 for the outdoor and without farm datasets, respectively). In the dataset including all sites, an IQR change in NO2 was associated with a 21% increase in NAL protein, which varied from 19% when adjusted for OPGSH(F) to 23% when adjusted for OPAA(C) or endotoxin (see online supplementary table S10). The significant positive association for absorbance in the outdoor dataset could not be disentangled from the association with NO2 because of the high correlation between them (0.74).

    Eighteen hours PE, there were no consistent positive associations. For water-soluble Fe, we observed a consistent negative association in the complete and without farm dataset (2 and 18 h PE). However, these associations became non-significantly positive after excluding data from the underground location (see online supplementary repository, RAPTES 2013).

    Associations with NAL lactoferrin

    In one-pollutant models in the complete and without farm datasets, we observed significant positive associations with all pollutants high at the underground location but not with others. No meaningful associations were found after excluding the underground site (see online supplementary repository, RAPTES 2013).

    Two-pollutant models were ineffective in identifying which of the pollutants high at the underground station had the strongest association: all significant associations observed in the one-pollutant models remained significant after adding a pollutant that was not high(est) at the underground, but decreased and became non-significant after adding another pollutant which was high at the underground site (summarised in table 4 for all sites, the full results of all three dataset are shown in see online supplementary tables S11–S13). As for the other biomarkers in NAL, the pattern of associations 18 h PE was generally similar, though weaker than 2 h PE (see online supplementary repository, RAPTES 2013).

    View this table:
    Table 4

    Associations between air pollution exposure and percentage changes (post–pre) in NAL lactoferrin 2 h after exposure in the dataset including all sites

    Associations with serum IL-6

    In one-pollutant models, we observed significant negative associations of IL-6 in serum with several PM characteristics, including endotoxin, OC(C) and water-soluble Cu, and O3. The only significant positive association we observed was for NO2, 18 h PE in the outdoor dataset, but this association reduced by over 50% and became non-significant after excluding 1% of the most influential observations (see online supplementary repository, RAPTES 2013).

    Results of two-pollutant models for serum IL-6 measured 2 h PE are shown in table 5 for all sites, and in online supplementary tables S1 4 and S15 for the outdoor and without farm datasets, respectively. Two-pollutant models for serum IL-6 measured 18 h PE are shown in the online supplementary repository (RAPTES 2013). In two-pollutant models, none of the exposure parameters were consistently significantly associated with changes in serum IL-6 at any time-point. In the complete dataset, there were however fairly consistent negative associations for OC(C) (2 and 18 h PE) and endotoxin (2 h PE), which were also associated with NAL IL-6, albeit in the opposite direction. Associations with endotoxin 2 h PE were similar but less consistent in the outdoor dataset, and not present in the without farm dataset. Effects of OC(C) 2 h PE were not present in the outdoor dataset, whereas in the without farm dataset effects were similar compared with the complete dataset but less consistent. Eighteen hours PE, the pattern of effects observed for OC(C) was comparable with the earlier time-point; however, these associations were reduced by approximately 50% and became non-significant after excluding 1% of the most influential observations (Cook's D analysis of one-pollutant models, see online supplementary repository, RAPTES 2013).

    View this table:
    Table 5

    Associations between air pollution exposure and percentage changes (post–pre) in serum IL-6 2 h after exposure in the complete dataset including all sites

    Discussion

    In the present study, we found associations between air pollution exposure and biomarkers for acute nasal inflammatory and cytotoxic responses in healthy subjects. The effects were largely driven by specific constituents rather than the total PM mass or OP as a suggestive integrated measure to predict health effects. The observed associations were stronger 2 h PE than 18 h PE. In two-pollutant models, consistent positive associations were found between NAL IL-6 and IL-8, with OC(C), NO3−, endotoxin and NO2; protein was associated with NO2; and NAL lactoferrin was associated with all PM characteristics that were substantially higher at the underground train station compared with the other sites. Unexpected negative associations were observed for IL-6 in blood with OC(C) and endotoxin.

    Although concentrations of OC(C) were the highest at the underground train station, associations between OC(C) and NAL cytokines disappeared after excluding the farm site while they were still present after excluding the underground site. OC originates from combustion processes as well as biogenic sources, for example, endotoxin in agricultural environments. Since OC(C) remained a significant predictor of NAL cytokines after adjusting for endotoxin, it is therefore most likely that the observed associations for OC(C) represent other bioaerosols related to farming such as fungi.21

    Consistent positive associations for NO3− with both NAL IL-6 and IL-8 as observed in the complete dataset disappeared after excluding the farm site. Since NO3− concentrations were the highest at the farm site, this might be another farm-related pollutant inducing nasal inflammation. However, there is limited toxicological evidence to support a causal association between NO3− and adverse health effects.22 When comparing the results of the complete and outdoor datasets, it appeared that the association for NO3− was stronger than for OC(C) in the complete dataset, whereas the opposite was observed in the outdoor dataset. Yet, as discussed above, the outdoor dataset is more relevant for analysing effects of farm-related OC(C). This led us to the conclusion that the observed associations for NO3− represent the effect of farm-related OC(C) rather than an individual effect of NO3−.

    In two-pollutant models, endotoxin remained a significant predictor of both NAL IL-6 and IL-8 controlling for all other pollutants. Effect estimates for endotoxin expressed per IQR were small (0.2%–0.6% increase in NAL cytokines per 0.2 EU/m3) and this would probably not reflect an adverse effect in healthy subjects.23 However, the farm site was not included in calculating the IQR. Levels at the farm site were more than 15 times higher (expressed as EU/m3) than at any other site.13 This indicates that \the magnitude of the effect would be much higher at outdoor locations with high endotoxin levels. Indoor levels of endotoxin in pig barns are known to be high, and have been linked with increased NAL cytokine levels24 ,25 and respiratory complaints.26 ,27 In the recent years, endotoxin exposure data on the general population have become available.28 ,29 However, studies on the relationship between outdoor endotoxin levels and health effects on the general population are lacking.

    We observed consistently significant positive associations for NO2 with NAL protein in all datasets, and with NAL cytokines in the without farm dataset. The fact that the latter associations were not consistently statistically significant in the complete and outdoor datasets indicates that the strong associations for endotoxin and OC hampered the identification of possible associations for other pollutants. Currently, there is no consensus whether the associations observed between NO2 and adverse health outcomes are due to direct effects of NO2 or due to other PM components covarying with NO2.30 Interestingly, in our data, changes in NAL cytokines were associated with NO2 after controlling for a large suite of PM characteristics including PNC, but not vice versa. This suggests that NO2 may have an independent effect on the biomarkers we studied.

    In one-pollutant models, increases in lactoferrin concentrations were associated with all PM characteristics that were substantially higher at the underground train station location (PM mass concentration, absorbance, EC, OC, metals and OP) compared with the outdoor sites. We hypothesised that changes in lactoferrin concentrations would be related to specific particle characteristics, namely, OP and iron or other transition metal content. However, individual effects of different pollutants could not be disentangled in the two-pollutant models. Ghio et al31 exposed human bronchial epithelial cells to ambient PM collected nearby a power plant and observed increased expression of lactoferrin on both mRNA as well as protein expression levels. They suggest that the observed increase could be attributed to the high particle metal content, since addition of the metal chelator deferoxamine significantly diminished lactoferrin synthesis and secretion. Furthermore, in vivo studies showed that intratracheal exposure to metal-rich particles enhanced lactoferrin production in both healthy rats and human subjects.31 ,32 In view of these results, a possible explanation for the increased NAL lactoferrin concentration observed in our study could be the high metal concentrations measured at the underground train station site.

    Remarkably, we found fairly consistent negative associations for OC(C) and endotoxin with serum IL-6. Since we also observed associations for OC(C) and endotoxin with NAL IL-6, albeit in the opposite direction, it could be a causal effect instead of a chance finding related to multiple testing. Barregard et al33 exposed healthy subjects to wood smoke and clean air for 4 h and found lower serum IL-6 levels after exposure to wood smoke. Similar effects were observed after exposure to concentrated ambient particles34 and ultrafine carbon particles.35 Frampton et al35 discussed a possible biological mechanism which involves penetration of ultrafine particles in the pulmonary capillary bed. This may in turn result in the release of anti-inflammatory cytokines that interact with the surface of particles and prevent or decrease an inflammatory response in the blood.

    Neither PM mass concentration nor particle OP was a consistent predictor for acute changes in nasal inflammatory markers in two-pollutant models. The latter was in contrast with our hypothesis, since we expected OP to be a better predictor than individual PM components. The absence of associations between the examined endpoints and OP may reflect the fact that the assay employed only examined the intrinsic potential of the particles to drive oxidation reactions in an acellular model reflecting their content of redox-active transition metals and quinones rather than upon interaction with a biological system.14 Since PM can elicit oxidative stress through alternative pathways upon interaction with the cellular/tissue matrix, an acellular assay does not necessarily reflect the total oxidative activity in vivo.7 Future studies should assess whether different in vitro models or in vivo OP assays could be better predictors for PM-induced health effects.

    To our best knowledge, there is only one other study that examined the effects of multiple ambient PM characteristics in association with biomarkers for acute nasal airway inflammation in schoolchildren.36 In one-pollutant models, they observed a 3.45% increase in percentage of neutrophils and a 29.98 pg/ml increase in IL-8 in NAL on the day of exposure per 11.5 µg/m3 change of PM2.5 and these effects remained significant in two-pollutant models. There were no consistently significant associations in two-pollutant models for any of the other pollutants (ie, PM2.5–10, O3, CO, NO2 and SO2). We, however, did not find significant associations with NAL IL-8 for PM2.5, but differences in study design, for example, study population and exposure time, hampered a comparison with our study.

    The strengths and limitations of our study design were discussed in detail previously.13 ,14 Briefly, using a semiexperimental design enabled us to define two-pollutant models to investigate independent effects of a large number of individual PM characteristics. Inevitably, some correlations between pollutants were still too high to interpret two-pollutant models. Another advantage of our study design was that exposure measurement error was small compared with studies relying on data from central monitoring sites.15 Since we specified a large number of models to investigate all possible combinations of air pollutants measured, we potentially induced chance findings in our results. We chose not to apply adjustments for multiple comparisons in our analyses as this is controversial in epidemiology.37 We recognise this may affect nominal levels of significance but surmise that, rather than performing adjustments, our findings call for replication in independent studies. In addition, we focused on the consistency of the associations rather than on individual significant results.

    Furthermore, NAL sampling is an easy to perform, non-invasive method to obtain biomarkers related to upper airway inflammation. Since we measured changes between PE and pre-exposure instead of single measurements for each subject, the interpretation of our results is not (as much) influenced by inter-individual variation as this is commonly relatively large for biomarkers measured in NAL. A disadvantage of using NAL is that, so far, no data are published on reference values in NAL, and the influence of using different NAL sampling techniques has not yet been systematically evaluated. In addition, there is limited information on the relationship between nasal inflammation and events occurring in the more distal airways.

    Conclusions

    We were able to identify specific PM characteristics associated with different biomarkers of acute pro-inflammatory effects in the nasal airways. In two-pollutant models, we found no consistent associations with OP and PM mass concentration. Instead, we observed consistent associations with nasal inflammatory markers for other PM characteristics, specifically OC, endotoxin and NO2.

    Acknowledgments

    The authors wish to thank all study participants; Daan Leseman, John Boere, Paul Fokkens, Marja Meijerink, Maartje Kleintjes, Veerle Huijgen, Jet Musters and Lise van den Burg for their excellent support in data collection; Miriam Groothoff, for medical supervision; and Eef van Otterloo for his help with participants’ recruitment.

    References

    1. ↵
      1. Brunekreef B
      . The color of smoke. Epidemiology 2010;21:903–4.
      OpenUrlCrossRefPubMedWeb of Science
    2. ↵
      WHO. Health relevance of particulate matter from various sources. Germany Copenhagen, Denmark: World Health Organisation Regional Office for Europe; 26–27 March 2007, Report on a WHO workshop Bonn.
    3. ↵
      1. Grahame TJ,
      2. Schlesinger RB
      . Health effects of airborne particulate matter: do we know enough to consider regulating specific particle types or sources? Inhal Toxicol 2007;19:457–81.
      OpenUrlCrossRefPubMedWeb of Science
    4. ↵
      1. Nel A
      . Atmosphere. Air pollution-related illness: effects of particles. Science 2005;308:804–6
      OpenUrlAbstract/FREE Full Text
    5. ↵
      1. Li N,
      2. Kim S,
      3. Wang M,
      4. et al
      . Use of a stratified oxidative stress model to study the biological effects of ambient concentrated and diesel exhaust particulate matter. Inhal Toxicol 2002;14:459–86.
      OpenUrlCrossRefPubMedWeb of Science
    6. ↵
      1. Borm PJ,
      2. Kelly F,
      3. Kunzli N,
      4. et al
      . Oxidant generation by particulate matter: from biologically effective dose to a promising, novel metric. Occup Environ Med 2007;64:73–4.
      OpenUrlFREE Full Text
    7. ↵
      1. Ayres JG,
      2. Borm P,
      3. Cassee FR,
      4. et al
      . Evaluating the toxicity of airborne particulate matter and nanoparticles by measuring oxidative stress potential—a workshop report and consensus statement. Inhal Toxicol 2008;20:75–99.
      OpenUrlCrossRefPubMed
    8. ↵
      ICRP. Human respiratory tract model for radiological protection. A report of a Task Group of the International Commission on Radiological Protection. Ann ICRP 1994;24:1–482.
      OpenUrlFREE Full Text
    9. ↵
      1. Kongerud J,
      2. Madden MC,
      3. Hazucha M,
      4. et al
      . Nasal responses in asthmatic and nonasthmatic subjects following exposure to diesel exhaust particles. Inhal Toxicol 2006;18:589–94
      OpenUrlCrossRefPubMed
    10. ↵
      1. Larsson BM,
      2. Grunewald J,
      3. Skold CM,
      4. et al
      . Limited airway effects in mild asthmatics after exposure to air pollution in a road tunnel. Respir Med 2010;104:1912–18.
      OpenUrlCrossRefPubMed
    11. ↵
      1. Mudway IS,
      2. Blomberg A,
      3. Frew AJ,
      4. et al
      . Antioxidant consumption and repletion kinetics in nasal lavage fluid following exposure of healthy human volunteers to ozone. Eur Respir J 1999;13:1429–38.
      OpenUrlAbstract/FREE Full Text
    12. ↵
      1. Steenhof M,
      2. Gosens I,
      3. Strak M,
      4. et al
      . In vitro toxicity of particulate matter (PM) collected at different sites in the Netherlands is associated with PM composition, size fraction and oxidative potential—the RAPTES project. Part Fibre Toxicol 2011;8:26.
      OpenUrlCrossRefPubMed
    13. ↵
      1. Strak M,
      2. Steenhof M,
      3. Godri KJ,
      4. et al
      . Variation in characteristics of ambient particulate matter at eight locations in the Netherlands—The RAPTES project. Atmos Environ 2011;45:4442–53.
      OpenUrlCrossRefWeb of Science
    14. ↵
      1. Strak M,
      2. Janssen NAH,
      3. Godri KJ,
      4. et al
      . Respiratory health effects of airborne particulate matter: the role of particle size, composition and oxidative potential—The RAPTES Project. Environ Health Perspect 2012;120:1183–9.
      OpenUrlCrossRefPubMedWeb of Science
    15. ↵
      Strak M, Hoek G, Steenhof M, et al. Components of ambient air pollution affect thrombin generation in healthy humans: the RAPTES project. Occup Environ Med 2013;70:33240.
      OpenUrlAbstract/FREE Full Text
    16. ↵
      1. Stenfors N,
      2. Nordenhall C,
      3. Salvi SS,
      4. et al
      . Different airway inflammatory responses in asthmatic and healthy humans exposed to diesel. Eur Respir J 2004;23:82–6.
      OpenUrlAbstract/FREE Full Text
    17. ↵
      1. Behndig AF,
      2. Mudway IS,
      3. Brown JL,
      4. et al
      . Airway antioxidant and inflammatory responses to diesel exhaust exposure in healthy humans. Eur Respir J 2006;27:359–65.
      OpenUrlAbstract/FREE Full Text
    18. ↵
      National Research Council. Biologic markers in pulmonary toxicology. Washington, DC: The National Academies Press, 1989.
    19. ↵
      1. Actor JK,
      2. Hwang SA,
      3. Kruzel ML
      . Lactoferrin as a natural immune modulator. Curr Pharm Des 2009;15:1956–73.
      OpenUrlCrossRefPubMed
    20. ↵
      1. Godri KJ,
      2. Duggan ST,
      3. Fuller GW,
      4. et al
      . Particulate matter oxidative potential from waste transfer station activity. Environ Health Perspect 2010;118:493–8.
      OpenUrlPubMedWeb of Science
    21. ↵
      1. Palmberg L,
      2. Larsson BM,
      3. Malmberg P,
      4. et al
      . Induction of IL-8 production in human alveolar macrophages and human bronchial epithelial cells in vitro by swine dust. Thorax 1998;53:260–4.
      OpenUrlAbstract/FREE Full Text
    22. ↵
      1. Schlesinger RB,
      2. Cassee F
      . Atmospheric secondary inorganic particulate matter: the toxicological perspective as a basis for health effects risk assessment. Inhal Toxicol 2003;15:197–235.
      OpenUrlCrossRefPubMedWeb of Science
    23. ↵
      Health Council of the Netherlands. Endotoxins. Health-based recommended occupational exposure limit. The Hague: Health Council of the Netherlands, 2010.
    24. ↵
      1. Ehnhage A,
      2. Kolbeck KG,
      3. Juto JE,
      4. et al
      . Swine dust exposure is a model for rapid induction of non-allergic neutrophil inflammation in the nasal mucosa of healthy volunteers, and the symptoms as well as the microcirculation are modified by nasal lavage. Rhinology 2007;45:292–8.
      OpenUrlPubMedWeb of Science
    25. ↵
      1. Wang Z,
      2. Larsson K,
      3. Palmberg L,
      4. et al
      . Inhalation of swine dust induces cytokine release in the upper and lower airways. Eur Respir J 1997;10:381–7.
      OpenUrlAbstract
    26. ↵
      1. Iversen M,
      2. Kirychuk S,
      3. Drost H,
      4. et al
      . Human health effects of dust exposure in animal confinement buildings. J Agric Saf Health 2000;6:283–8.
      OpenUrlCrossRefPubMed
    27. ↵
      1. Vogelzang PF,
      2. van der Gulden JW,
      3. Folgering H,
      4. et al
      . Endotoxin exposure as a major determinant of lung function decline in pig farmers. Am J Respir Crit Care Med 1998;157:15–18.
      OpenUrlCrossRefPubMedWeb of Science
    28. ↵
      1. Mueller-Anneling L,
      2. Avol E,
      3. Peters JM,
      4. et al
      . Ambient endotoxin concentrations in PM10 from Southern California. Environ Health Perspect 2004;112: 583–8.
      OpenUrlCrossRefPubMedWeb of Science
    29. ↵
      1. Schulze A,
      2. van Strien R,
      3. Ehrenstein V,
      4. et al
      . Ambient endotoxin level in an area with intensive livestock production. Ann Agric Environ Med 2006;13:87–91.
      OpenUrlPubMedWeb of Science
    30. ↵
      WHO. Air quality guidelines. Global Update 2005. Particulate matter, ozone, nitrogen dioxide, and sulfar dioxide. Copenhagen, Denmark: World Health Organisation Regional Office for Europe, 2006.
    31. ↵
      1. Ghio AJ,
      2. Carter JD,
      3. Samet JM,
      4. et al
      . Metal-dependent expression of ferritin and lactoferrin by respiratory epithelial cells. Am J Physiol 1998;274:L728–36.
      OpenUrlWeb of Science
    32. ↵
      1. Ghio AJ,
      2. Richards JH,
      3. Dittrich KL,
      4. et al
      . Metal storage and transport proteins increase after exposure of the rat lung to an air pollution particle. Toxicol Pathol 1998;26:388–94.
      OpenUrlAbstract/FREE Full Text
    33. ↵
      1. Barregard L,
      2. Sallsten G,
      3. Gustafson P,
      4. et al
      . Experimental exposure to wood-smoke particles in healthy humans: effects on markers of inflammation, coagulation, and lipid peroxidation. Inhal Toxicol 2006;18:845–53.
      OpenUrlCrossRefPubMedWeb of Science
    34. ↵
      1. Gong H Jr.,
      2. Linn WS,
      3. Sioutas C,
      4. et al
      . Controlled exposures of healthy and asthmatic volunteers to concentrated ambient fine particles in Los Angeles. Inhal Toxicol 2003;15:305–2.
      OpenUrlCrossRefPubMedWeb of Science
    35. ↵
      Frampton MW, Utell MJ, Zareba W, et al. Effects of exposure to ultrafine carbon particles in healthy subjects and subjects with asthma. Res Rep Health Eff Inst 2004;126:1–47; discussion 49–63.
    36. ↵
      1. Chen BY,
      2. Chan CC,
      3. Lee CT,
      4. et al
      . The association of ambient air pollution with airway inflammation in schoolchildren. Am J Epidemiol 2012;175:764–74.
      OpenUrlAbstract/FREE Full Text
    37. ↵
      1. Streiner DL,
      2. Norman GR
      . Correction for multiple testingcorrection for multiple testingIs there a resolution? CHEST J 2011;140:16–18.
      OpenUrlCrossRefWeb of Science

    Supplementary materials

    • Supplementary Data

      This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

      Files in this Data Supplement:

    Footnotes

    • Contributors MSte, ISM, IG, GH, KJG, FJK, RMH, RHHP, FRC, EL, BB, MStr and NAHJ were involved in conception and design of the study. MSte and MStr collected the samples and data. MSte and MStr cleaned and statistically analysed the data. MSte, GH, BB and NAHJ interpreted the data. MSte, MStr and KJG performed laboratory analyses. MSte, BB and NAHJ drafted the paper. MSte, ISM, IG, GH, KJG, FJK, RMH, RHHP, FRC, EL, BB, MStr and NAHJ revised the draft paper.

    • Funding The RAPTES project was funded by the RIVM Strategic Research Program (S630002).

    • Competing interests None.

    • Patient consent Obtained.

    • Ethics approval Ethics approval was provided by METC Utrecht.

    • Provenance and peer review Not commissioned; externally peer reviewed.

    • Data sharing statement All one- and two-pollutant data can be found in a web only file (see online supplementary repository, RAPTES http://www.iras.uu.nl/rsc/repository/steenhof-et-al-oem2013-online-repository-raptes.pdf) which will be published on our institute's website (open access).