Air Pollution, Airway Inflammation, and Lung Function in a Cohort Study of Mexico City Schoolchildren

Background The biological mechanisms involved in inflammatory response to air pollution are not clearly understood. Objective In this study we assessed the association of short-term air pollutant exposure with inflammatory markers and lung function. Methods We studied a cohort of 158 asthmatic and 50 nonasthmatic school-age children, followed an average of 22 weeks. We conducted spirometric tests, measurements of fractional exhaled nitric oxide (FeNO), interleukin-8 (IL-8) in nasal lavage, and pH of exhaled breath condensate every 15 days during follow-up. Data were analyzed using linear mixed-effects models. Results An increase of 17.5 μg/m3 in the 8-hr moving average of PM2.5 levels (interquartile range) was associated with a 1.08-ppb increase in FeNO [95% confidence interval (CI), 1.01–1.16] and a 1.07-pg/mL increase in IL-8 (95% CI 0.98–1.19) in asthmatic children and a 1.16 pg/ml increase in IL-8 (95% CI, 1.00–1.36) in nonasthmatic children. The 5-day accumulated average of exposure to particulate matter < 2.5 μm in aerodynamic diamter (PM2.5) was significantly inversely associated with forced expiratory volume in 1 sec (FEV1) (p = 0.048) and forced vital capacity (FVC) (p = 0.012) in asthmatic children and with FVC (p = 0.021) in nonasthmatic children. FeNO and FEV1 were inversely associated (p = 0.005) in asthmatic children. Conclusions Exposure to PM2.5 resulted in acute airway inflammation and decrease in lung function in both asthmatic and nonasthmatic children.

VOLUME 116 | NUMBER 6 | June 2008 • Environmental Health Perspectives Research | Children's Health Exposure to air pollution has been associated with decrements in lung function (Castillejos et al. 1995;Gauderman et al. 2007;Moshammer et al. 2006;Romieu et al. 1997Romieu et al. , 1998Svendsen et al. 2007) and an increase in respiratory symptoms (Castillejos et al. 1992;Leonardi et al. 2000), effects to which asthmatic children appear more susceptible (Romieu et al. 1996;Ward and Ayres 2004). Experimental and epidemiologic studies have also shown that fossil fuel-combustion products act as an adjuvant in the immune system and may lead to enhancement of allergic reaction (Nel et al. 1998) to airway inflammation in susceptible subjects (Holguin et al. 2007;Leonardi et al. 2000;Yeatts et al. 2007) with potential long-term effects (Calderon-Garcidueñas et al. 2006). However, few longitudinal epidemiologic studies have evaluated the impact of fossil fuel combustion on airway inflammatory response (Fischer et al. 2002;Koenig et al. 2003Koenig et al. , 2005Steerenberg et al. 2001) using an integrated approach including markers of airway inflammatory response and lung function changes.
Asthma is a complex disease characterized by inflammation and hyperresponsiveness of the airways. Nitric oxide signaling pathways have been implicated in the regulation of airway hyperresponsiveness and recently of fractional exhaled nitric oxide (Fe NO ) (Van Amsterdam et al. 2000) and the pH of exhaled breath condensate (EBC) samples has been proposed as a noninvasive biomarker to assess airway inflammation (Rosias et al. 2004). Exhaled NO is produced endogenously in the airways from L-arginine by NO synthase (NOS) (Silkoff et al. 2000), and two constitutive isoforms and cytokineinduced overexpression of NOS are likely to contribute to its increase (Wu et al. 2007). Similarly, increased cytokine levels have been described in asthmatic patients and interleukin (IL)-8 may contribute to inflammatory response in the airways (Narayanan et al. 1999).
The aim of the present study was to evaluate the association of air pollutant and traffic-related exposures with Fe NO levels, IL-8 levels in nasal lavage, pH of EBC, and lung function in a cohort of school-age asthmatic and nonasthmatic children. We hypothesized that short-term exposure to air pollution and traffic-related emissions was associated with increased airway inflammation and decreased lung function.

Material and Methods
Study design. A dynamic panel study of asthmatic and nonasthmatic schoolchildren living in Mexico City was conducted during June 2003-June 2005. The children were enrolled during the first 10 months of the study (June 2003-April 2004 and followed for an average of 22 weeks. Spirometric tests and measurements of Fe NO levels, IL-8 levels in nasal lavage, and pH of EBC were conducted during the follow-up. Location and population. The study population consisted of children living in three Mexico City municipalities, Iztapalapa, Iztacalco, and Netzahualcóyotl, where high levels of traffic-related emissions are the major source of pollutants. The Mexican Ministry of Transport and Communications (Secretaría de Transportes y Vialidad, Mexico City, Mexico) reports that on weekdays about 35,000 vehicles pass through the intersection of two highways in one direction, and that about 2,000 diesel-fueled vehicles per day use one of the highways in the study area (Gobierno del Distrito Federal 1999). Similarly, the traffic count conducted in this area indicated a flow of between 303 and 1,840 vehicles per hour (Romieu et al. 2008a).
One hundred fifty-eight asthmatic children attending the Hospital Infantil de Mexico Federico Gómez, one of the largest pediatric hospitals in the city (Mexico City), were invited to participate in the study. The diagnosis and severity of their asthma were based on clinical symptoms and response to treatment and rated by a pediatric allergist as mild (intermittent or persistent), moderate, or severe according to the Global Initiative for Asthma (GINA) guidelines (Global Initiative for Asthma 2006). Fifty nonasthmatic children were recruited by asking the asthmatic children to invite a schoolmate or a friend from their neighborhood. The children in both groups were between 6 and 14 years of age. They lived in the study area, attended public schools located close to their home, were volunteers, and were not selected using probability-based sampling. Figure 1 presents the study area, showing schools, monitoring stations, major highways, and the location of the children's homes. All procedures were explained to the parents, who signed an informed consent form. The children also gave their informed assent. The study protocol was reviewed and approved by ethics committees at both the National Institute of Public Health and the Hospital Infantil de Mexico.
Collection of health outcomes. We collected data using a general-purpose questionnaire (adapted from existing survey instruments) on sociodemographic variables, past health history, and potential indoor environmental exposures (tobacco smoke and pets in the home). Information on allergy test results, medication, and medical visits in the preceding 2 years was obtained from the medical record. At baseline and every 15 days during follow-up, a respiratory symptoms questionnaire was applied, anthropometric measurements were taken, spirometric tests performed, Fe NO levels measured, and samples of nasal lavage and EBC obtained. Spirometric tests were performed on 158 asthmatic children (1,503 measurements) and 50 nonasthmatic children (591 measurements). Measurements were repeated an average of 11 times (range, 5-21) per subject during the study period. We were unable to take all samples for inflammatory markers from all the children, for logistic reasons. We obtained 702 measurements of Fe NO from 126 asthmatic children and 302 from 50 nonasthmatic children, 759 measurements of IL-8 in nasal lavage from 129 asthmatic children and 285 from 45 nonasthmatic children, and 551 measurements of pH in exhaled breath condensate from 119 asthmatic children and 201 from 44 nonasthmatic children.
Spirometry. The spirometric tests were performed according to American Thoracic Society (ATS) specifications (ATS 1995) using an EasyOne spirometer (ndd Medical Technologies, Andover, MA, USA). The tests were conducted in a room with stable temperature and relative humidity. All lung function tests were performed by the same technicians, and the best of three technically acceptable tests was selected.
Exhaled nitric oxide determination. The levels of Fe NO were measured following the ATS guidelines (ATS 1999) during outpatient visits to a clinic. Children were seated for at least 5 min before commencing the measurement and throughout the procedure; all measurements were conducted indoors to minimize inhaled NO-free external air. NO was measured by chemiluminescence, using a continuous analyzer (CDL 88 sq; ECO Physics, Ann Arbor, MI, USA). The Fe NO reading was displayed on the monitoring system and the mean of three acceptable tests was taken.
Nasal lavage. Nasal lavage was performed following the methodology proposed by Diaz-Sanchez et al. (1994), with the subject sitting with the nasopharynx closed while tilting the neck back 45°from the vertical. Five milliliters of warm (37°C) normal saline is instilled into each nostril by pipette. After 10 sec, during which the subject shakes the head softly from side to side, the subject brings the head forward, expelling the wash fluid into a plastic receptacle. The subject then performs up to four further nasal washes at 30-sec intervals, with each wash being collected in a separate tube. We measured different cytokines including IL-8, interferon gamma, IL-6, and IL-10 levels in nasal lavage in the laboratory of D. Diaz-Sanchez, using commercially available ELISA kits according to the manufacturer's instructions. However, except for IL-8, the levels in most of the samples were below the detection limit and we report only the IL-8 results. For logistic reasons, we did not determine cellular composition.
EBC collection. EBC was collected using an R-tube, and the breath was cooled by placing an aluminum cooling sleeve over the disposable polypropylene tube (Hunt 2002). Samples were obtained following the ATS/ERS (European Respiratory Society) Task Force recommendations (Horvath et al. 2005;Hunt 2002). Participants were asked to breathe tidally through the mouthpiece connected to the R-tube for 10 min to collect approximately 2 mL of exhaled breath fluid, which was aliquoted and frozen to -70°C within 15 min of collection.
Exposure assessment. Exposure was estimated from outdoor PM 2.5 (particulate matter < 2.5 µm in aerodynamic diameter), NO 2 and O 3 concentrations recorded by the Mexico City government at four fixed-site central monitoring [Red Automática de Monitoreo Afmosférico (RAMA)] locations within the study area (Cerro de la Estrella and Hangares; Merced; Universidad Autonoma Metropolitana, Iztapalapa; and La Perla). Daily average, maximum moving average and 8-hr maximum ozone, nitrogen dioxide, and PM 2.5 concentrations, and meteorologic data (temperature and humidity) were obtained for all (505) days of the study period. The home of each participating child was georeferenced using a geographic information system (GIS), and the closest monitoring station was assigned to the child. All children attended public schools located close to their home, and no fixed-site monitoring station was > 5 km from a child's home or school. We also conducted monitoring at each school for three 15-day periods during the follow-up to validate data obtained from the fixed-site monitoring stations (RAMA). Local daily 24-hr average PM 2.5 was determined using Mini-Vol portable air samplers (version 4.2;  Airmetrics, Eugene, Oregon, USA) with 47-mm Teflon filters (R2PJ047; Pall Gelman, Ann Arbor, MI, USA) and flows set at 5 L/min, and 7-day integrated data for NO 2 and O 3 concentrations were obtained using Ogawa passive samplers (Ogawa USA, Pompano Beach, FL, USA) (Watson et al. 1999). Samplers were located outside the 37 schools, generally on the roof, at a height of up to 4 m and far from any objects (e.g., trees, buildings) that would prevent air flow. Gravimetric analysis of the 47-mm Teflon filters was performed at the air laboratory of the National Center for Environmental Research and Training (CENICA) in Mexico City. The NO 2 and O 3 filters were assembled at the Mexico City laboratory. After exposure, all badges were placed in sealed bags and sent to the Harvard School of Public Health for chemical analysis (Levy et al. 1998). Statistical analysis. The basic characteristics of the two groups of children were compared by bivariate analysis using the t-test, the Fisher exact test, or the chi-square test, depending on variable type. The short-term association of traffic-related pollutants (PM 2.5 , O 3 , and NO 2 concentrations at the fixed-site monitoring stations) with the health outcomes was studied using linear mixed-effects models, considering models for continuous and binary response. This enabled us to appreciate the variability within and between subjects. We ran models with both random intercept and random slope, and with random intercept only. Because the coefficients were similar in both types of model, we present results from the linear mixed model with random intercept only. The model is as follows: where X i = Z i X i is the appropriate (n i x p) matrix of known covariates with fixed effects β and subject-specific effects b i and ε i in an n i -dimentional vector of residual components. Another advantage of the models used is that they do not discard subjects with incomplete data. We determined the goodness of fit of each model using residual diagnosis and the Hausman specification test (Hausman 1978). Based on our previous work and international studies conducted among asthmatic children (Delfino et al. 2004(Delfino et al. , 2006Romieu et al. 1996Romieu et al. , 2002Romieu et al. , 2004Schildcrout et al. 2006), we hypothesized a priori that the acute effect of air pollutants on pulmonary functions would occur with lags of 1-2 days and that a larger effect would be observed when considering cumulative exposure over several (up to 5) days, whereas for inflammatory markers (Fe NO levels, IL-8 levels, pH of EBC) we hypothesized a shorter response with 0-to 1-day lag. We modeled several pollution exposure indices (8-hr maximum moving average, 24-hr average, 24-hr maximum average, accumulated days). Pearson correlations were determined between air pollutant levels and various climatic variables. Fe NO and IL-8 levels were not normally distributed and were log transformed. Models were adjusted for potential confounding factors including sex, body mass index, previous day minimum temperature, corticoid use, and chronological time. Other variables such as age, socioeconomic index (considering mother education and school type), outdoor activities, atopic status, exposure to environmental tobacco smoke, use of antiallergy medicine, and season were not significant (p > 0.10) and did not alter the results by >1%. Analyses were conducted using STATA (version 9.2; StataCorp., College Station, TX, USA). Table 1 presents the characteristics of the study population. The median age of participants was 9.6 years [quartile (Q) 25: 7.9; Q75: 11.0) for the asthmatic and 9.3 (Q25: 7.9; Q75: 11.5) for the nonasthmatic children. Of the asthmatic children, 55% were classified as having mild intermittent, 26.9 % as having mild persistent, and 17.5% as having moderate persistent asthma, according to the GINA guidelines. Of the asthmatic children, 6% used an inhaled corticosteroid and 10% were prescribed antibiotics on at least one occasion during follow-up. Eighty-nine percent of the asthmatic children and 72% of the nonasthmatic children had positive skin prick tests. The most common sensitivities were to house dust mite (Dermatophagoides pteronyssinus), cat (Fel d 1), and cockroach (Blatella americana) allergens. Participants with asthma had higher average levels of Fe NO (p = 0.001) and lower average levels of IL-8 (p = 0.002). The pH measurements in EBC were similar in both groups (p = 0.632).

Results
Environmental exposure data. The 8-hr moving average PM 2.5 ranged from 4.24 to 102.8 µg/m 3 during the study period, with a mean of 28.9 µg/m 3 . It exceeded 30 µg/m 3 on 52% of the days (Figure 2). The 8-hr moving average NO 2 levels ranged from 14.9 to 77.6 ppb with a mean of 37.4 ppb. The 8-hr moving average O 3 level ranged from 4.9 to 86.3 ppb with a mean of 31.6 ppb ( Table 2). The correlation between PM 2.5 and O 3 was r = 0.46 (p = 0.000). The correlations between O 3 and NO 2 and NO 2 and PM 2.5 were r = 0.28 (p = 0.000) and r = 0.61 (p = 0.000) respectively. Local measurements conducted at the children's schools were correlated with levels at the central monitoring stations (r = 0.77 for PM 2.5 , r = 0.21 for NO 2 , and r = 0.60 for O 3 ). The mean (± SD) of local measurements was 26.3 ± 12.5 µg/m 3 for PM 2.5 , 35.05 ± 12.6 ppb for NO 2 , and 26.9 ± 9.5 ppb for O 3 .
Association between health outcomes and air pollutants. The associations between traffic-related pollutants and main outcomes are shown in Table 3 for asthmatic children and  Table 4 for nonasthmatic children. After adjusting for sex, body mass index, corticoid use, previous day minimum temperature, and chronological time, an increase of 17.5 µg/m 3 in the 8-hr moving average of PM 2.5 levels [equivalent to one interquartile range (IQR)] was associated with a 1.08-ppb increase in Fe NO [95% confidence interval (CI), 1.01-1.16] in asthmatic children, whereas in nonasthmatic children the association was marginally significant (p < 0.081). In contrast, an increase of 17.5 µg/m 3 in 8-hr moving average PM 2.5 levels (IQR) was more strongly associated with IL-8 levels in nonasthmatic children (1.16 pg/mL increase; 95% CI, 1.00-1.36) than in asthmatic children (1.08 pg/mL increase; 95% CI, 0.98-1.19). Similarly, the 8-hr moving average of NO 2 was associated with an increase in Fe NO in asthmatic children but not in nonasthmatic children, whereas it was associated with IL-8 in nonasthmatic children but not in asthmatic children. The 8-hr moving average of O 3 was associated with both Fe NO and IL-8 in asthmatic children. The pH of EBC was significantly associated with O 3 exposure only in asthmatic children. Further adjustment for the use of allergic medicine did not modify these results.
Cumulative PM 2.5 exposure over the 5 preceding days was associated with a significant decrement in forced vital capacity (FVC) in both asthmatic and nonasthmatic children, whereas the decrement in FEV 1 was significant only in asthmatic children (Figure 3). For FEV 25-75 (the middle 25-75% of the forced expiratory volume measurement), the associations were not significant (Tables 3  and 4). Various models were run considering different lags; however, the effect of PM 2.5 on FVC and FEV 1 (FEV in 1 sec) was stronger for a 1-day and a 5-day lag (data not shown). No significant associations between NO 2 and O 3 and lung function tests were observed in asthmatic or nonasthmatic children.
Among asthmatic children, the increase in Fe NO was associated with a decrease in FEV 1 (p = 0.005), particularly among those with mild intermittent and moderate asthma. IL-8 was also inversely associated with FEV 1 but the association did not reach significance (p = 0.18).
In multipollutant models including O 3 as well as PM 2.5 , we noted that among asthmatic children O 3 remained significantly related to inflammatory markers (Fe NO , IL-8, and pH of EBC) whereas PM 2.5 remained inversely related to pulmonary functions (FEV 1 and FVC) but lost its significant effect on Fe NO . Among nonasthmatic children, the inverse relation between PM 2.5 and FEV 1 and FVC persisted, whereas PM 2.5 and O 3 had a lesser effect on inflammatory markers (data not shown).
A large proportion (72%) of the nonasthmatic children were atopic, so we repeated the analysis including only atopic children without asthma. In this subgroup we observed a significant increase of Fe NO related to PM 2.5 exposure, suggesting that atopy increases the Fe NO response (data not shown).

Discussion
The results of the present cohort study of asthmatic and nonasthmatic children show that Fe NO levels, IL-8 levels in nasal lavage, pH of EBC, and changes in lung function are associated with acute exposure to trafficrelated air pollutants. Changes in Fe NO were inversely associated with FEV 1 in asthmatic children, suggesting that the inflammatory response of airways most likely influences the decrease in lung function. The effects on inflammatory markers were higher for PM 2.5 and O 3 concentrations than for NO 2 . The effect appears on the same day as the exposure and can cumulate over several days, resulting in lung function decrement after 4 or 5 days of cumulative exposure.
Although exposure to outdoor ambient levels of PM 2.5 , NO 2 , and O 3 has been associated with increased asthma and respiratory symptoms in children, there are few previous longitudinal studies combining inflammatory response and change in lung function in response to air pollution, particularly in nonasthmatic children. Epidemiologic panel studies report that exposure to air pollutants increases the Fe NO levels in susceptible children. Delfino et al. (2006), showed that an IQR increase of 73 µg/m 3 in 1-hr maximum personal PM 2.5 was associated with a 0.60-ppb increase in Fe NO (95% CI, 0.14-1.05) in asthmatic children who live in an urban area. A panel study of asthmatic children in Seattle, Washington, reported that ambient PM 2.5 concentrations were also associated with increased levels of exhaled NO . We recently reported an increase in Fe NO in asthmatic children living close to roads (50-m buffer) whereas no effect was observed in nonasthmatic children; but no data on lung function or other inflammatory markers were available (Fischer et al. 2002;Holguin et al. 2007;Steerenberg et al. 1999). The present study provides evidence of the adverse effect of traffic-related pollution not only in asthmatic but also in nonasthmatic children.
The effect of PM 2.5 on IL-8 was stronger in nonasthmatic than in asthmatic children. This was contrary to our expectation. However this has already been reported in another population, in which exposure to diesel exhaust provoked airway inflammation with airway neutrophilia and an increase of IL-8 in healthy subjects shortly after exposure but did not induce neutrophilic response in asthmatic subjects (Stenfors et al. 2004). A potential explanation is that, in asthmatics, IL-10 might inhibit the synthesis of many inflammatory cytokines including IL-8, which has low baseline levels in asthmatic subjects (Barnes and Lim 1998;Stenfors et al. 2004). However, our data are inconclusive on this point. Our interest in estimating the impact of air pollution on IL-8 response is based on the fact that, though not a key player in asthmatic response, its control through a redox-sensitive promoter makes it a good marker of inflammatory response linked to air pollutants (Ichinose et al. 1997). Oxidative effects can increase secretions of proinflammatory cytokines, chemokines, and adhesion molecules and enhance allergic reaction in the airway (Behndig et al. 2006;Takizawa 2004). In particular, oxidative stress has been shown to up-regulate expression of the gene that encodes IL-8 in a variety of cells (Narayanan et al. 1999), leading to the recruitment of neutrophils to sites of inflammation (Diaz-Sanchez and Riedl 2005;Romieu et al. 2008b). In our study, O 3 (a very potent oxidant) was associated with strong inflammatory responses in both asthmatic and nonasthmatic children.
One difference between asthmatic and nonasthmatic children is the impact of air pollutant exposure on Fe NO . Exhaled NO is a sensitive biomarker of airway inflammation (Dinakar 2004); it has been shown to be increased in asthmatic and atopic subjects (Barnes 1996;Gaston et al. 1994;Stamler et al. 2001;Van Amsterdam 2000) and to be a sensitive marker for the respiratory effects of air pollution (Fischer et al. 2002;Holguin et al. 2007). In our study, asthmatic children had a higher baseline Fe NO level than nonasthmatic children, and nonasthmatic children who were atopic had a higher Fe NO baseline level than those who were not [13.9 ppb (IQR = 33.4) vs. 7.7 ppb (IQR = 8.5)]. A significant increase in Fe NO after exposure to air pollutants was observed among the asthmatic children, though no significant response was observed in the nonasthmatic children. However, when we analyzed nonasthmatic atopic children separately, we observed a significant increase of Fe NO after PM 2.5 exposure, suggesting that atopic children are more responsive to exposure than nonatopic children.
Another contribution of our study is the evaluation of pH of EBC related to air pollutant exposure. We hypothesized that asthmatic children would have a lower exhaled breath pH at baseline than nonasthmatic children and that exposure to air pollutants would lead to a decrease in the pH of exhaled condensate as a marker of lower airway inflammation. The exhaled breath condensate of asthmatics contains a high concentration of reactive oxygen species (ROS) and reactive nitrogen species (RNS), reflecting changes in the lower respiratory tract during inflammation and affecting local pH. This acidification might contribute to asthma pathophysiology (Hunt et al. 2000). Exposure to air pollutants, particularly O 3 , which has a strong oxidative potential and acts in the lower airway, is likely to result in an increase in ROS and RNS. This would affect the pH of the lower respiratory tract, reflecting local inflammation. We found that the pH of EBC decreased in both nonasthmatic and asthmatic children with air pollutant exposure, and significantly so with O 3 exposure in asthmatic children. Brunetti et al. (2006) observed that the pH of EBC in patients with acute asthma was lower than that of patients with stable asthma, and the amplitude of the effect we observed among asthmatic children in response to O 3 (16-ppb increase) was similar to that described by these authors. Our results therefore add a plausible VOLUME 116 | NUMBER 6 | June 2008 • Environmental Health Perspectives   biological mechanism to the association of traffic-related emissions and adverse respiratory health effects. An important issue is the interrelation between inflammatory response and change in lung function in relation to air pollution exposure. Among asthmatic children, FEV 1 was inversely related to Fe NO , IL-8, and PM 2.5 exposure, but the coefficient of PM 2.5 decreased when the inflammatory markers were introduced in the models. In contrast, among nonasthmatic children this adjustment strengthened the inverse association between FEV 1 and PM 2.5. This suggests that, in asthmatic children, the inflammatory response may partly explain the effect of PM 2.5 on FEV 1 . We also observed a significant inverse association between decrement in FEV 1 and increase in Fe NO in asthmatic children. However, given the complexity of inflammatory responses to air pollutants, these results need to be interpreted with caution.
Some limitations must be taken into account when interpreting the results of this study. Daily variations in air pollutant exposure were evaluated through the daily records of the fixed central monitoring locations (RAMA). The temporal variations in each child's exposure were assumed to follow those at the central monitoring site. To strengthen the validity of this assumption, each child was assigned to the monitoring site closest to his or her home by means of a spatial GIS, providing greater variability in the data. We also monitored PM 2.5 , NO 2 , and O 3 at the schools and correlated the data from the two sets of monitors. The correlations between the fixed central monitors and the local monitors were 0.77, 0.21, and 0.60 respectively. The lowest correlation was for NO 2 , as in other studies (Kodama et al. 2002), whereas the highest was for PM 2.5 , which might explain why significant changes in lung function were observed only for this pollutant, possibly because of better estimation of exposure. However, specific components of PM 2.5 also appear to play a role in the adverse effect of this pollutant. When we analyzed data on elemental carbon levels measured in a subsample of PM 2.5 filters (n = 37) from the schoolbased monitors, we observed that elemental carbon levels were inversely associated with FEV 1 and FVC in a subgroup of 30 asthmatic children (data not shown) (p = 0.005), highlighting the role of heavy vehicle (diesel) emissions as a source of exposure (Janssen et al. 2001) and strengthening the causal relationship with PM 2.5 . Finally, although some degree of exposure may be confounded by other unmeasured factors such as variations in socioeconomic status, this is unlikely in the present study because all our participants came from the same study area and attended the same public school system; and further adjustment by socioeconomic status did not modify our results.
Most of the asthmatic children in this study were classified as having mild intermittent asthma (55%) and only 17.5% as having moderate persistent asthma. This limited our ability to study the impact of air pollution on asthma severity. In previous studies conducted in Mexico City, we had observed that children with moderate to severe asthma were more susceptible to air pollution (Romieu et al. 2004). Our results might therefore underestimate the impact of air pollution on children with more severe asthma.

Conclusion
Our data show that Fe NO levels, IL-8 levels in nasal lavage, pH of EBC, and changes in lung function are associated with acute exposure to traffic-related air pollutants. These adverse effects were observed in a longitudinal setting in a free-living population, more specifically in a cohort of schoolchildren including nonasthmatic children. Among the asthmatic children, air pollution exposure was related to an increase in Fe NO levels and a decrease in lung function, whereas neutrophil airway inflammation as well as a decrease in lung function was observed in the nonasthmatic children. These results could have significant public health policy implications, because a large proportion of schools in Mexico City and other countries are located very close to roads with heavy traffic.