Ozone Exposure and Lung Function^*: Effect Modified by Obesity and Airways Hyperresponsiveness in the VA Normative Aging Study FREE TO VIEW

Many studies and metaanalyses¹²³ have found associations between acute ambient ozone exposure and increased risk of death. In addition, ambient ozone has been associated with increased respiratory-related emergency department visits and hospital admission in the elderly.⁴⁵⁶ These findings suggest that the elderly are particularly susceptible to the effects of ozone.

Lung function, as measured by spirometry, has been shown to predict all-cause mortality and cause-specific (such as respiratory and cardiovascular) mortality.⁷⁸⁹ Exposure to ozone has been documented to cause decrements in lung function in healthy subjects.¹⁰¹¹ By identifying the characteristics of people most susceptible to the effects of ozone on lung function, we may gain insight into subpopulations most at risk of other ozone-induced effects on morbidity and mortality, as well as into the mechanisms responsible for the ozone effects.

Few studies have examined the lung function response to ozone in the elderly. Many studies have been published examining younger populations, but since susceptibility to the effects of ozone has been shown to vary with age,¹² the effects of ozone in the elderly are still largely unknown. Past studies,^13141516 which typically include only healthy nonsmokers, have found that functional responsiveness to ozone has been no greater, and usually lower, among older adults than in young adults. Unexplained intersubject variability has been observed in responsiveness to ozone as measured by lung function,¹⁷¹⁸ and research to identify sources of this variation is needed, especially in the elderly who are especially susceptible to the morbidity and mortality associated with acute ozone exposure.

Airway hyperresponsiveness (AHR), a central feature of asthma, is associated with both low levels of lung function and faster longitudinal lung function decline.¹⁹ Asthmatics appear to be a subset of subjects who are at higher risk for the effects of ozone. For example, exacerbations and emergency department visits for asthma have been associated with ozone exposure.²⁰²¹ It appears that adverse effects of ozone in asthmatics are mediated by effects of ozone on airway inflammation²² and subsequent AHR.²³

The role of obesity in responsiveness to ozone has not previously been studied in humans but has been documented in mice. These studies have found that acute ozone exposure results in increased AHR and inflammation in obese mice compared to nonobese control mice,²⁴ and those effects do not depend on the mode of obesity.²⁵ These results in mice suggest that obesity may be a modifying factor for the response of humans to acute ozone exposure.

In this longitudinal epidemiologic study, we examine two possible modifiers of the effect of acute exposure to ambient ozone on lung function. We hypothesized that obesity and AHR would each modify the effect of ozone on lung function. We also investigated whether there is evidence for interaction between obesity and AHR in modification of lung function response.

Subjects in this study were part of the Veterans Administration Normative Aging Study,²⁶ a longitudinal study established in 1963, details of which have been published previously. Briefly, the study enrolled 2,280 men from the Greater Boston area, ages 21 to 80 years, who were determined to be free of known chronic medical conditions by an initial health screening. Participants visited the study center repeatedly to undergo physical examinations, including pulmonary function testing, and fill out questionnaires approximately every 3 years. We included the subset of 904 subjects whose lung function was measured between January 1995 and June 2005 and whose AHR was measured during the same visit or at a prior visit. This study was approved by the Institutional Review Boards of all participating institutions, and informed consent was obtained from all subjects.

Study center visits took place in the morning after an overnight fast and abstinence from smoking. Physical examinations included measurement of height, weight, lung function (FVC, FEV₁), and methacholine challenge testing. Subjects filled out questionnaires on smoking habits and pulmonary disorders (asthma, chronic bronchitis, emphysema) based on the American Thoracic Society-Division of Lung Diseases-1978 questionnaire.²⁷ Descriptive statistics for these data are listed in Table 1 .

Pulmonary function and methacholine challenge tests were performed as previously reported.²⁸ Briefly, measures of FVC and FEV₁ were obtained using a water-filled recording spirometer, and values were adjusted by body temperature and pressure. Spirometric tests were performed in accordance with American Thoracic Society guidelines.

Methacholine challenge tests were performed using a procedure adapted from that of Chatham and colleagues.²⁹ Methacholine inhalations were administered at incremental doses corresponding to 0, 0.330, 1.98, 8.58, 16.8, and 49.8 μmol. For our analysis, we defined a positive response to methacholine challenge (AHR) when a subject experienced a 20% decline in FEV₁ following any of the doses at or before 8.58 μmol. Subjects whose FEV₁ did not decline by 20% in response to any of the administered doses and subjects who demonstrated a 20% decline in FEV₁ only from a higher methacholine dosage (16.8 or 49.8 μmol) were categorized as having no AHR response.

Methacholine challenge tests were administered between 1984 and 2000. We used data from the most recent methacholine challenge test available for each subject at that visit. Subjects were excluded from methacholine challenge testing for having ischemic heart disease or baseline FEV₁ at < 60% of the predicted value, and some subjects elected not to participate.

At each visit, height and weight were collected and used to calculate the body mass index (BMI). Obesity was defined as a BMI ≥ 30 kg/m², consistent with the definition given by the National Institutes of Health.³⁰

Four monitoring sites measured ambient ozone continuously, with one monitor located in each of four cities in the Greater Boston area: Boston, Chelsea, Lynn, and Waltham. All monitors conformed to US Environmental Protection Agency standards. In our analyses, we used the average of the measurements from each monitor. Although the ozone measurements spanned 10 years, we did not create long-term averages of ambient ozone because all subjects lived in the same metropolitan area and thus the long-term averages would be the same for each subject. Instead, since subjects visited on different days throughout the year, we calculated short-term averages of the ambient ozone concentration a few days before the visit time of that subject and analyzed the data as a repeated-measures study. Since subjects visited on different days, the short-term exposure to ambient ozone before lung function measurement varied from subject to subject. In a previous study, we examined the effect of ozone on lung function using averages ranging from 1 to 5 days prior to the study visit and found that the 48-h average was most significantly associated with decline in both FVC and FEV₁ (S.E. Alexeeff, BSc; unpublished data; November 2006). Hence, we used the average ozone concentration 48 h prior to the study visit for the analyses of effect modification presented here.

We obtained meteorologic data including temperature and relative humidity from the Boston airport weather station. To control for outdoor weather, we used the apparent temperature, defined as a person’s perceived air temperature.³¹³² We calculated the average apparent temperature for the 48 h before the study visit to correspond with the 48-h ozone concentration average. Descriptive statistics for apparent temperature and ozone concentration are listed in Table 2 .

Ozone concentrations had low correlations with particulate matter < 2.5 μm (0.29), carbon monoxide (−0.21), and nitrogen dioxide (−0.15), and hence we believed it was reasonable to examine ozone alone without adjusting for other pollutants. We have limited data on particle components, but for the days we do have data (approximately one third), ozone correlations were as expected: ozone was not correlated with elemental carbon (0.05) but had a moderate correlation with sulfates (0.45).

Lung function measurements FVC and FEV₁ were log-transformed to increase normality and stabilize variance. We chose the following variables a priori and included them in our model: age, height, race, cigarette smoking (smoking status and pack-years), chronic lung conditions (asthma, emphysema, and chronic bronchitis), season, year of visit, weekday, apparent temperature, and ozone concentration.

We measured the FVC and FEV₁ of each subject on up to four visits, with a mean of 2.3 visits per subject. We used a mixed model for our regression analyses to account for the repeated measurements on the same subjects. This type of model allows each subject to act as his/her own control, which accounts for intrasubject variability. Hence, in addition to using the variables listed above to account for factors known to affect lung function, our longitudinal design also allows us to control for unmeasured within-subject variation by using a random intercept for each person. An association between the dependent variable and a covariate was considered to be significant if the covariate had a p value < 0.05 in the model.

We tested the modification effects of AHR and obesity by using an interaction term with ozone. We report the significance of the interaction term and also the effect estimate (and confidence limit) within each category (eg, obese and not obese). We present the estimated effect of ozone as a percentage change in lung function. Because FVC and FEV₁ were log_e-transformed in our model, the percentage changes in FVC and FEV₁ were calculated by [exp(ΔO₃ × β)-1] × 100%, with 95% confidence intervals (CIs) [exp(ΔO₃ × [β ± 1.96 × SE]) − 1] × 100%, where exp(X) is the mathematical constant e (≈ 2.718) raised to the power of X, ΔO₃ is the change in ozone concentration, β is the estimated regression coefficient, and SE is the SE of β. For the change in ozone concentration, we used the increment of 15 parts per billion (ppb), which is approximately equal to the interquartile range of the 2-day ozone concentration averages.

To test the linearity of the association and the possible existence of a threshold, we reran our mixed-effects model using the mixed-model formulation of a penalized spline for ozone concentration. This allowed us to specify a nonlinear regression spline with up to five degrees of freedom for the ozone dose response. The coefficients representing the nonlinear components of the ozone dose-response curve are treated as random, and estimated using restricted maximum likelihood.³³

When we fit the spline model for ozone, the restricted maximum likelihood estimation chose one degree of freedom (linear) as the best-fitting model of spline models with up to five degrees of freedom. That is, the random coefficients representing the deviation from linearity were estimated as zero. This indicates that a linear dose response is the best fit across the range of exposure in our study.

To evaluate the modifying effects of AHR, subjects were classified in categories of “response” and “no response” for a methacholine dosage ≤ 8.58 μmol, with a response defined in the “Materials and Methods” section above. For a 15-ppb increase in ozone, greater decreases in FVC and FEV₁ were estimated for those in the response group compared to the no-response group (Table 3 , Fig 1, 2 ). This interaction between ozone and AHR was significant for predicting FEV₁ (p = 0.044) but not significant for FVC. Note that while the CIs for these two groups are overlapping (Fig 1), this effect is indeed significant, and it has been shown in general that an interaction effect can be significant while CIs have some overlap.³⁴³⁵

To assess of the effect of obesity as a modifier of response to ozone, we estimated the effect of ozone for the two categories “obese” and “not obese” (categories explained above). For a 15-ppb increase in ozone, the obese were estimated to have greater drops in both FVC and FEV₁ than the nonobese (Table 3, Fig 1, 2). This interaction between ozone and obesity was significant when predicting FEV₁ (p = 0.022) but not significant for FVC.

We also investigated the combined effect of obesity with AHR by separating the cohort into four classes (class 1A, not obese with no AHR; class 1B, not obese with AHR; class 2A, obese with no AHR; and class 2B, obese with AHR). Results are listed in Table 3 and shown in Figures 1, 2. For both FVC and FEV₁, the estimated decrease in lung function associated with ozone exposure was greater in classes 1B and 2A than in class 1A, with the most severe estimates given for class 2B.

The results of the interaction model suggested a trend where the effects increased in magnitude across classes 1A through 2B, so we performed a trend test. We were interested in testing the hypothesis that the effect of obesity and AHR would have a more than additive effect. Because we had no prior hypothesis as to whether obesity or AHR would result in a greater effect modification for ozone, we defined our trend variable as the interaction of a 1,2 variable for obesity with a 1,2 variable for AHR. Our resulting nonlinear trend was 1,2,4, where the presence of either risk factor alone was coded 2. The trend was significant for both FVC (p = 0.048) and FEV₁ (p < 0.001).

Our analyses support our hypothesis that obesity and AHR each modify the effect of ozone on lung function. We found that a greater lung function decline in response to acute ozone exposure is predicted for those who are obese compared to the nonobese, and for those with AHR compared to those without AHR. In addition, we found evidence of an interaction between AHR and obesity that also modifies the decrease in lung function from ozone exposure. We also found that the response to ozone appears to be linear for the range of exposures in our study, indicating the presence of a response at lower exposure levels.

Obesity has not been studied in humans as a modifier of the effect of responsiveness to ozone. However, our results are consistent with studies²⁴²⁵ of mice, which have found that responses to ozone (inflammation and increased AHR) were greater in obese mice compared to nonobese control mice. Our findings showing greater decreases in lung function in obese humans compared to the nonobese lend further evidence to indicate that obesity is an important factor in responsiveness to ozone.

Our results showing greater decreases in lung function in those with AHR are consistent with the findings of increased responsiveness to ozone in those with asthma. Asthmatics have been shown to exhibit greater declines in lung function in response to ozone than nonasthmatics.³⁶ Responsiveness to ozone as measured by inflammatory markers has also been shown to be increased in asthmatics compared to those without asthma.³⁷

In our examination of obesity together with AHR, our findings suggest that obesity and AHR interact. This interaction seems reasonable considering that other studies have linked obesity to AHR and asthma. We previously identified a relationship between obesity and AHR, including a finding that those with a “high” BMI (> 29.4 kg/m²) were associated with the development of AHR.³⁸ Additionally, a large body of epidemiologic data, including both cross-sectional and prospective studies,³⁹⁴⁰ suggests that obesity is a risk factor for asthma. Thus, the many studies that have found associations between obesity and AHR or between obesity and asthma lend support to our hypothesis that AHR and obesity may interact in modifying a person’s response to ozone.

A possible mechanism to explain our findings of the effects of obesity and AHR in ozone responsiveness involves lung inflammation, which is widely known to be an effect of ozone. How closely ozone-induced decline in lung function is related to this inflammation is still unknown. Some controlled studies⁴¹⁴² examining the effects of ozone found both increased levels inflammation and decrements in FVC and/or FEV₁ after exposure, but found no correlation between the magnitudes of these responses. However, fibrinogen concentration has been correlated with the magnitudes of decrements in forced expiratory flow, midexpiratory range.⁴³

Inflammation, as measured by blood eosinophil count, has been associated with responsiveness to methacholine in this cohort.⁴⁴ Airway inflammation is thought to play a key role in the development of asthma,⁴⁵ and the inflammatory response to acute ozone exposure has been shown to be increased in asthmatics compared to normal subjects.³⁷ Also, systemic inflammation is strongly associated with obesity in both children and adults.⁴⁶ Since AHR and obesity are both associated with increased inflammation, we suspect that those with either trait are more likely to be susceptible to the inflammatory effects of ozone, leading to lower FVC and FEV₁. Additionally, it seems plausible that those who have both obesity and AHR experience further increased inflammation, and therefore are even more prone to the effects of ozone, which would explain the interaction seen in our results.

In addition to inflammatory responses, other mechanisms have been identified that could explain the link between obesity and asthma, including mechanical factors like the stretching of muscles in the airway and levels of various adipokines (proteins synthesized and released from adipose tissue), including hormones involved in antiinflammatory responses and energy regulation.⁴⁷ Given the mounting evidence of a relationship between obesity and asthma, our finding that obesity and AHR appear to interact in response to ozone seems reasonable. In terms of our finding that the size of the interaction between AHR and obesity appears to be more than additive, it is not clear which of the possible mechanisms is most likely responsible. We considered whether obesity may simply be a marker for more severe cases of AHR. After comparing the methacholine dose-response slopes for the obese and not obese who had AHR, we found no difference between the slopes (results not shown). Thus, we believe that obesity is not simply a marker for extreme AHR and is instead having some other effect that is interacting with AHR. As the mechanisms linking obesity to asthma and AHR are further investigated, reason for a greater than additive effect of these factors may be identified.

Our analysis of the linearity of the response to ozone indicates that there is no evidence of threshold across the range of exposures in our study. We cannot exclude the possibility of a threshold at lower concentration, but since the tenth percentile of 2-day ozone concentration was only 11 ppb, we expect that any threshold would exist only at a very low exposure level. In addition, the US standard of 80 ppb for an 8-h peak average was exceeded on only 23 days (2.3% of total days used in our study). In conjunction with the evidence of the linearity of the response to ozone, this suggests that the elderly experience acute lung function decrements in response to exposure levels below the US standard.

A limitation of this study is that the study population consists of only elderly men, most of whom are white. These results cannot easily be extrapolated across age or gender because ozone responsiveness as measured by lung function varies with age and possibly also with gender.¹² Also, lung function is known to vary with race,⁴⁸⁴⁹ and lung function response to ozone may also vary with race.⁵⁰ Given these differences in responsiveness, we cannot predict the degree to which these factors might modify the response to ozone in other populations. We do suspect, however, that obesity and AHR modify the effect of ozone response in other populations at some level, and this points to an important area for future research.

Our study is also limited by relying on ambient exposure without accounting for individual exposure using activity patterns. While the lack of activity pattern data in our study is a concern, we have no reason to believe this could upwardly bias our effect, and the likely direction of the bias is downward.⁵¹ Still, small studies⁵²⁵³ of personal exposure have demonstrated moderate to low associations between ambient ozone levels and personal ozone exposure and some correlation with personal exposure to other products of photochemistry, including particulates, with effects varying by season. The only copollutant with a moderate correlation to ozone was sulfate, so we do not know whether some of the effect seen could be attributed to a combination of ozone and sulfates. Additional research is needed to address the association of personal exposure to ozone and other pollutants with pulmonary responses in large longitudinal studies.

This study provides evidence of modification of the lung function response to acute ozone exposure in the elderly by both obesity and AHR. The evidence for a possible interaction between obesity and AHR that modifies ozone response suggests the importance of examining combinations of risk factors to help explain the intersubject variability in responsiveness to ozone .

We thank Elaine R. Dibbs, Shelly Amberg, and Jordan Awerbach for their invaluable contributions to the Veterans Administration Normative Aging Study.