We found suggestive evidence that the acute effects of ozone exposure on asthmatic children’s pulmonary function may depend on both antioxidant genes and dietary intake of vitamin C. These ozone effects appear more pronounced in persistent asthmatic children. In addition, the cumulative effect of genetic susceptibility in GSTM1, GSTP1, and NQO1 polymorphisms and low intake of vitamin C was associated with the largest decrement in FEF25-75 among children with persistent asthma.
GSTM1 has been most intensively studied as a deleted versus a nondeleted dichotomous polymorphism; there are few data using GSTM1 copy number. In this study, we found a potential dose–response effect of GSTM1 on the relationship between FEF25-75 and ozone exposure. This result is consistent with the dose–response relationship between GSTM1 and the risk of lung cancer on smokers
In human lung cells, GSTP1 Val/Val genotype has been associated with reduced GST Pi enzymatic activity
. Our finding that the Val/Val may confer higher susceptibility to ozone damage than Ile/Val or Ile/Ile genotypes is consistent with that result. In addition, some epidemiologic studies have reported associations between Val/Val genotype and reduced lung function in young adults
 and increased risk of early onset of asthma
[25, 26]. Discrepancies with other studies
[27, 28] that report the Val allele as protective may be related to ethnic differences among populations, with their consequent differences in linkage disequilibrium patterns, or due to change variation among studies.
Regarding diet, persistent asthmatics with low daily vitamin C intake had significant decrements in FEF25–75 associated to ozone exposure, but the effect of ozone was lessened among children with vitamin C intake over 105 mg/day. Our results suggest that persistent asthmatic children exposed to ozone might benefit from vitamin C intake above the RDI. Intake of > 105 mg/day was reached in our study from diet alone. This suggests that children do not need to take supplements. Consumption of daily fruits and vegetables may provide enough vitamin C for lessening the effects of ozone, on lung function, regardless the combination of antioxidant genotypes. Furthermore, fruits and vegetables provide additional nutrients that supplements do not include.
When genetic and dietary data were combined, persistent asthmatics with both low vitamin C intake and GSTM1 null or Val/Val GSTP1 had, on average, lower levels of FEF25–75 than other children.
Although our selection of 105 mg/day as the cutoff was based on the data distribution, it is according with the increment of 100 mg/d suggested by some studies for increasing about 10-50 ml in FEV1.
Using our genotype score as an index of genetic susceptibility, we found that the effect of ozone on persistent asthmatics with less than 105 mg/day vitamin C and with 4 to 6 risk alleles was slightly stronger than that estimated for GSTM1 under the single gene approach in the same dietary group (−97.2 ml/s against −91.2 ml/s respectively). The difference between both effects was not significant (p = 0.88). This result suggests that the GSTM1 genotype alone could have been used as an indicator of susceptibility to ozone.
As a result of environmental health policies in Mexico City, concentrations of ozone have been reduced over time. From 1998 to 2000 (first cohort enrollment), the average concentrations were between 100 and 125 ppb; while by the second cohort enrollment (2003–2005), concentrations were between 85 to 100 ppb
. When we fit separated models by cohort, we found that in the first cohort, the effect of ozone on children with low vitamin C intake and GSTM1 null genotype, was −96.0; 95% CI(−204.0,12.6); whereas a lower effect was observed in the second cohort (−20.4; 95% CI (−114.6, 73.2)). Thus, final models were adjusted for a covariate accounting for the cohort.
Although our focus was on FEF25-75, we also analyzed the association between ozone and other measures of lung function including FEV1, FVC and PEF but we did not find any significant effect.
Regarding to the assessment of the potential confounding role of ethnicity, we rejected the null hypothesis of association between Native-American ancestry and the outcome setting an alpha-level of 0.20 instead of 0.05 level in order to insure adequate power to detect any important confounder effect
. This lack of confounder effect may be explained by the fact that the children were recruited at the same Public Hospital suggesting similar socioeconomic status.
Some advantages and limitations of our study need to be addressed. Advantages include repeated measurements of FEF25-75. Obtaining longitudinal measurements of quantitative phenotypes reduces potential outcome misclassification and increases power by focusing on within-subject variations in outcome while controlling for among subjects differences using random intercepts and random slopes on time-dependent ozone exposure.
We were able to assess the potential confounding effect of population admixture and thereby exclude any potential confounding effect of admixture from our data. This is the first time that the confounding role of ethnicity in the association between genotype and respiratory health of Mexican asthmatic children exposed to ozone has been assessed. Ethnicity was based on genome-wide association genotyping. The absence of association between the levels of FEF25-75 and ethnicity in these Mexican asthmatic children supports the validity of previously-reported genetic associations in this population
Potential limitations include inaccuracy in assessment of personal ozone exposure. Because exposure assessment was based on monitoring stations located within 5 km from the children’s homes, potential misclassification of exposure is possible. However, using personal monitors on a subset of participants in the “Antioxidants” study (n = 144), a significant association (p < 0.01) between personal and ambient ozone levels was observed
. Based on this validation, we would expect any misclassification in exposure to be random with the consequence of underestimating the adverse effect of ozone.
We were not able to measure plasma vitamin C levels of the participants but a validated food frequency questionnaire was used to estimate intake. Self-reported vitamin C intake was adjusted for total caloric intake. Through this adjustment, the between-person variation due to over- or under reporting of intake is reduced and a gain in accuracy is potentially obtained
[18, 34]. Under this assumption, we would expect any misclassification in vitamin C intake to be random; therefore its potential protective effect may be underestimated.
While we focused on dietary vitamin C intake, we recognize that vitamin E has also been positively related to lung function. Children in the highest level of vitamin C intake had 50 mg per day of vitamin E as part of the antioxidant supplementation. It is worthy to note that although children in the medium level of vitamin C did not have supplementation, they had no significant ozone effects. Therefore, the effect of vitamin C might be predominant. Nevertheless, since both vitamin C and E may be present in the same foods, and because the biological interplay between them
, part of the effect observed for vitamin C may be related to vitamin E intake as well.
Although ozone has received a great deal of attention, PM2.5 has also been associated with respiratory diseases. However, since concentrations of PM2.5 in México City only began to be registered by the network in 2000, we were not able to analyze the effect of this pollutant. Nevertheless, when data were available, it has been observed that PM2.5 and ozone are correlated (r = 0.46; p < 0.0001)
. Thus, part of the effect observed for ozone may be related to PM2.5 exposure.
Selection bias is unlikely because those children without diet or genotype information were not aware of their individual susceptibility to ozone exposure. In addition, included and excluded groups had similar characteristics. Moreover, we used an objective measure of ozone effect – pulmonary function.
We acknowledge that our sample size was relatively small. Large sample sizes are required to study multi-way interactions with complex diseases such as asthma. In contrast, longitudinal studies of quantitative traits with repeated measurements are more powerful than cross-sectional studies of binary disease outcomes. Thus, despite our modest sample size, we were able to examine effects of ozone and diet on repeated measures analyses of the quantitative traits of FEF25-75 in our asthmatic children.
Although we are not aware of a study with similar design that could be used for replication, a number of observational studies suggest that vitamin C may reduce asthma risk and may prevent inflammatory response in the airways by reducing reactive oxygen species
. Further, the intake of vitamin C and of fruits rich in vitamin C has been positively associated to lung function
. Moreover, the protective effect of high fruits and vegetables intake has been reported as having potentially inverse modifying influence to cigarette smoking on the asthma risk for genetically vulnerable individuals
 and the antioxidant role of some genes such as GSTM1, GSTP1, and NQO1 has been recognized
[6, 7, 26].
Albeit reported p-values throughout this paper have not been corrected for multiple comparisons, we believe that it is unlikely to detect effects due to statistical fluctuations only because of the epidemiological evidence for the interactive effects of air pollutants with vitamin C intake
[29, 35, 38] and antioxidant genes
[25, 27, 28] on asthma symptoms and lung function.