CC16 Levels into Adult Life Are Associated with Nitrogen Dioxide Exposure at Birth

Abstract

Rationale: Lung function and growth are adversely associated with nitrogen dioxide (NO₂) exposure. Lower levels of circulating club cell secretory protein (CC16) in childhood are also associated with subsequent decreased lung function. NO₂ exposure may induce epithelial damage in lungs and alter club cell proliferation and morphology.

Objectives: To determine if increased ambient NO₂ levels at participants’ home addresses in early life were associated with decreased levels of CC16 from age 6 to 32 years.

Methods: Participants were enrolled at birth in the Tucson Children’s Respiratory Study and had circulating CC16 measured at least once between age 6 and 32. Linear mixed models were used to determine the association between estimated ambient NO₂ exposure at participants’ home address at birth or age 6 with CC16 levels from age 6 to 32.

Measurements and Main Results: NO₂ exposures at birth or age 6 were available for 777 children with one or more CC16 measurement. We found a negative association between NO₂ exposure and CC16 levels, with a 4.7% (95% confidence interval, −8.6 to −0.7) decrease in CC16 levels from age 6 to 32 per interquartile range increase in NO₂ exposure (6.0 ppb) at the participants’ birth address. We observed modification by race (p interaction = 0.04), with stronger associations among participants with at least one black parent (−29.6% [95% confidence interval, −42.9% to −13.2%] per interquartile range). NO₂ at participant’s age 6 address was not significantly associated with CC16 levels (−1.9%; 95% confidence interval, −6.3 to 2.6).

Conclusions: Higher exposure to NO₂ at birth is associated with persistently low levels of CC16 from 6 to 32 years.

At a Glance Commentary

Scientific Knowledge on the Subject

Lower levels of circulating club cell secretory protein in childhood are associated with reduced lung function in adulthood. Increased exposure to nitrogen dioxide is also associated with reduced lung function growth in children. Evidence from in vitro and in vivo studies suggests that nitrogen dioxide exposure may induce epithelial damage in the lungs and alter club cell growth, yet no previous publication has assessed the effects of nitrogen dioxide exposure in children on subsequent circulating club cell secretory protein levels.

What This Study Adds to the Field

We used data from a longitudinal birth cohort study to determine the associations between nitrogen dioxide exposures in early life and circulating club cell secretory protein from age 6 to 32 years. We determined that higher exposures to nitrogen dioxide at the birth address are associated with persistently lower levels of circulating club cell secretory protein from childhood (age 6) into adult life (age 32). There was no association between nitrogen dioxide exposures at age 6 with circulating club cell secretory protein levels.

Nitrogen dioxide (NO₂) is a highly reactive oxidizing gas that arises from the reaction of nitric oxide and oxygen in the atmosphere (1). Both NO₂ and nitric oxide are by-products of fossil fuel combustion and are among the most common pollutants measured in both outdoor and indoor air (2). Primary outdoor sources include diesel vehicles and power plants, whereas common indoor sources are the pilot light for gas cook stoves and hot water heaters. NO₂ may be the best proxy pollutant to assess spatial exposure to the urban air pollution mixture (3), and nonwhites in the United States are significantly more likely to live in areas with increased NO₂ exposure (4).

Deficits in children’s lung function and growth are associated with exposure to NO₂ in both cross-sectional and prospective studies (5–7). A 20-year-long study found that long-term reductions in levels of NO₂ were associated with significant improvements in children’s lung function growth (8). Multiple studies document that inhaled NO₂, because of its reactive nature, can induce epithelial injury in the lungs (1, 9, 10). Of particular concern is evidence that during periods of lung development, epithelial cells may be uniquely susceptible to NO₂-induced death (1). Furthermore, NO₂ exposure is associated with altered club (formerly Clara) cell proliferation and morphology in animal studies (9).

These club cells release a secretory protein, referred to as CC16, which is a homodimeric pneumoprotein with antiinflammatory properties that may protect the lungs from oxidative stress (11–13). Because CC16 measured in serum has been correlated with levels in BAL fluid and club cell density, serum CC16 is proposed as a novel biomarker for early respiratory impairment from both acute and/or chronic lung injury (11, 14). Decreased levels of circulating CC16 are prospectively associated with lung cancer mortality, chronic obstructive pulmonary disease development, and lung function deficits in children and adults (15–17).

We postulated that early life exposure to NO₂ may be associated with lung damage, which is reflected in lower CC16 levels. To test this hypothesis, we used data from the Tucson Children’s Respiratory Study (CRS), a longitudinal birth cohort study that enrolled newborns to examine risk factors for acute lower respiratory tract illnesses in early childhood and chronic lung disorders later in life (18, 19). We determined whether ambient NO₂ levels at participants’ birth and age 6 addresses were associated with levels of CC16 from age 6 to 32 years. If early CC16 levels can be shown to reflect lung injury from NO₂ exposure, early life interventions that target children with low CC16 levels could be developed to prevent potential long-term health effects in high-risk children. Some of these results have been previously reported in the form of an abstract (20).

Methods

Study Population

The CRS is a longitudinal nonselected birth cohort study. All healthy infants born to mothers using the services of the largest health maintenance organization between 1980 and 1984 were eligible to participate and were contacted by a study nurse. Of the 1,596 eligible families contacted, 1,246 infants (78%) were enrolled. A detailed description of the cohort is published (18, 19). For the purpose of the current analysis, demographic characteristics of the child and parents, and home address, were obtained from a questionnaire administered at enrollment (within 2 wk after birth). Blood samples were obtained at age 6 (mean = 6.1; SD = 0.8), 11 (mean = 10.8; SD = 0.6), 16 (mean = 16.6; SD = 0.6), 22 (mean = 22; SD = 0.9), 26 (mean = 26.4; SD = 0.8), and 32 years of age (mean = 31.8; SD = 1.0). The study was approved by the University of Arizona Human Subjects Committee. Informed consent was obtained from a parent of a participating child and/or from the participants themselves starting at age 16 years.

CC16 Measurements

Circulating CC16 was measured in stored samples from each follow-up period. Serum samples were used for all time points, with the exception of 78 plasma samples from Year 6. In sensitivity analyses we excluded the plasma samples. Measurement techniques and the adjustments for plasma samples have been described in detail previously (16). All samples were cryopreserved at −80°C and circulating CC16 was measured using a commercially available ELISA kit (kit range: 1.57–50.0 ng/ml; BioVendor). All samples had detectable values.

NO₂ Exposure Assessment

We estimated NO₂ exposure at each enrollment (1980–1984) and age 6 (1986–1990) address using a land-use regression (LUR) model developed from NO₂ measurements obtained in eastern Pima County (where Tucson is located) between 1987 and 1991. NO₂ concentrations were measured in residential yards using passive Palmes diffusion tubes (21). Participating homes were from the Pima County Workers Study, who were recruited through county government departments with a wide range of occupations, resulting in a study cohort with varied socioeconomic status, demographic indicators, and geographic distribution. In 1987, the year closest to CRS enrollment, 166 homes were sampled for one to two 1-week periods, resulting in 311 observations.

We developed and evaluated our NO₂ LUR by adapting protocols for the European Study of Cohorts for Air Pollution Effects (ESCAPE), which has been conducted in 36 study areas in Europe (22). We created 89 geographic predictor variables based on ESCAPE protocols (22). Additional variables important to the Tucson environment were included (e.g., distance to airports, active mining operations, and agricultural land). Following the ESCAPE method, simple regressions were performed with all predictors and the one with the highest R² and slope in the predefined direction (e.g., area of nearby natural space is negatively related to air pollution) became the “seed” model. Other predictors were added one-by-one and kept if: 1) the adjusted R² increased by more than 1%, 2) the slope met the predefined direction, and 3) the slope direction of other predictors did not change. After all predictors were evaluated, variables in the model were removed one-by-one if their P value was greater than 0.1, beginning with the variable with the largest P value. Model performance was evaluated using a leave-one-out cross-validation method. Details on the source data and calculations for each variable are provided in online supplement (see Table E1 in the online supplement).

Using ArcGIS 10.1 (ESRI), we geocoded the enrollment and age 6 address for each participant. These coordinates were used as receptor points in the LUR to estimate ambient NO₂ exposure in ppb at each address as:

$$\left[\text{N}{\text{O}}_2\right]=6.18+\left(8.48\times {10}^{-6}\right)\text{P}\text{o}{\text{p}}_{5\text{k}\text{m}}-0.16\sqrt{\text{E}\text{l}\text{e}\text{v}}+247\left(\frac{1}{\text{D}\text{i}\text{s}{\text{t}}_{\text{a}\text{i}\text{r}}}\right)+67.6\left(\frac{1}{\text{D}\text{i}\text{s}{\text{t}}_{\text{r}\text{a}\text{i}\text{l}}}\right),$$

where Pop_5km is the population density within 5 km (persons/m²); Elev is the elevation (m); and Dist_air and Dist_rail are the distance to the nearest airport and rail line, respectively (m). To predict exposures corresponding to each participant’s birth or age 6 address, these variables were obtained for each address corresponding to the calendar year of enrollment (1980–1984) or age 6 (1984–1992) from the sources listed in Table E1, as in previous ESCAPE studies. Estimates were scaled based on differences between 1987 and the exposure year in NO₂ measurements from the regional monitoring network (7).

Statistical Analysis

All analyses were performed with R (V3.3.4; R Foundation for Statistical Computing). NO₂ exposures were scaled by dividing by the interquartile range to aid interpretability of the coefficients (i.e., a 1-unit increase is equal to the interquartile range). Because the distribution of CC16 values was right skewed, and concentrations are theoretically log-normally distributed (23), values were natural log-transformed. We used Pearson chi-square test to assess if the demographic characteristics of those who were included in the analyses were different from participants who were excluded. Simple linear regression was used to determine if there were associations between NO₂ exposures and participant characteristics.

To estimate longitudinal associations between NO₂ exposure and CC16 levels throughout life, we used linear mixed models and included participants with complete covariate and exposure data, and with CC16 measurements for at least one time point. Results are presented as percent change in CC16 per interquartile range increase in NO₂. Percent change was calculated from the β coefficients according to methods for log-transformed outcome variables (24). To select covariates for inclusion, we identified covariates a priori that are important in the air pollution–lung function literature, and examined associations between NO₂ exposure and those covariates (maternal education, maternal age, race/ethnicity, delivery method, marital status, maternal smoking, child sex, home heating, socioeconomic status, birth order, and parity) and covariates previously associated with CC16 in our cohort (25) (maternal age, sex, parental education, maternal smoking at age 6, CC16 genotype, race/ethnicity). We excluded variables that were strongly associated with the exposure but not with the outcome to avoid inducing instrumental variable bias (26). The remaining variables were included if they resulted in a change-in-estimate of the NO₂-CC16 relationship by greater than 10% or if they enhanced the SE of the estimate.

Final model covariates included fixed effects for child race/ethnicity, child sex, maternal smoking at enrollment, follow-up visit (6–32 yr), and random effects by subject. Unadjusted models omitted the fixed effects (with the exception of follow-up visit) but adjusted for random effects by subject. We examined associations between NO₂ exposure at birth and at age 6 separately, and CC16 levels at age 6, 11, 16, 22, 26, and 32 years. We assessed modification by several variables including race/ethnicity (non-Hispanic white, any Hispanic parent, any black parent, other), maternal education (high school or less vs. greater than high school), maternal smoking, child sex, and indices of neighborhood socioeconomic status and housing conditions derived from the U.S. census (27), with an alpha cutoff of 0.05 on the interaction term, using a likelihood ratio test. To assess if later CC16 measurements were associated with active smoking, we conducted a sensitivity analysis with active smoking between age 18 and 32 years.

Results

Of the original 1,246 children enrolled, 1,137 had enrollment and 806 children had age 6 addresses, for which NO₂ exposure could be estimated and were included in the analyses (Table 1). NO₂ values were estimated from the LUR models as described and reflect estimates at the child’s address during each time period. NO₂ levels were slightly higher during the enrollment period than during the first follow-up period (geometric means birth vs. age 6; P = 0.05). At enrollment, children classified as “Other” race/ethnicity had significantly lower exposures at their home address than children of any other race/ethnicity category (see Table E2). There were no other significant differences in NO₂ exposure at enrollment by participant characteristics. However, at age 6, children who had a Hispanic or black parent had higher NO₂ exposure levels, as did children with younger mothers or those whose mothers had lower educational attainment.

Table 1.

Distribution of Ambient NO₂ Concentrations (ppb) at Reported Residence at Enrollment and First Follow-up (Age 6)

	n	GM	GSD	Median	IQR	Range
Enrollment (1980–1984)	1,137	10.96	1.41	11.00	6.00	4.18–27.30
First follow-up (1986–1990)	806	10.23	1.48	10.74	6.69	2.73–20.39

Definition of abbreviations: GM = geometric mean; GSD = geometric standard deviation; IQR = interquartile range; NO₂ = nitrogen dioxide.

Characteristics and CC16 levels for the study population included in the analysis are presented by follow-up period in Table 2. As previously reported, CC16 levels in this cohort increased with age (25). There were no large differences in demographic characteristics of the participants by follow-up period (Table 2). However, CRS participants included in this analysis were more likely to be non-Hispanic white or Hispanic and have mothers with more education and that were less likely to smoke than those who were not included in the analysis (see Table E3).

Table 2.

Characteristics of Participants Included in Analyses in Each Follow-up Period

	Enrollment	6 yr	11 yr	16 yr	22 yr	26 yr	32 yr
n	1,137*	479†	550†	432†	416†	358†	234†
CC16, ng/ml, GM (GSD)		7.9 (1.5)	7.5 (1.5)	8.4 (1.5)	11.5 (1.5)	9.5 (1.6)	10.5 (1.5)
NO2 at enrollment, ppb, median (IQR)	11.0 (6.0)	10.8 (6.3)	10.9 (6.2)	10.7 (6.2)	10.8 (6.2)	10.7 (6.0)	10.8 (6.2)
NO2 at age 6, ppb, median (IQR)		10.5 (6.5)	10.8 (6.5)	10.7 (6.4)	10.5 (6.4)	10.3 (6.4)	10.4 (6.9)
Child sex, % (n)
M	48.3 (550)	49.9 (239)	48.7 (268)	48.8 (211)	48.5 (204)	47.2 (169)	43.6 (102)
F	51.6 (587)	50.1 (240)	51.3 (282)	51.2 (221)	51.5 (217)	52.8 (189)	56.4 (132)
Child ethnicity race, % (n)
Non-Hispanic white	59.4 (675)	62.2 (298)	58.6 (322)	60.2 (260)	61.3 (255)	62.0 (222)	60.3 (141)
Any Hispanic	25.5 (290)	27.8 (133)	31.5 (173)	29.6 (128)	29.6 (123)	28.5 (102)	30.3 (71)
Any black	4.1 (47)	3.6 (17)	3.3 (18)	3.2 (14)	2.2 (9)	2.5 (9)	2.6 (6)
Other	11.0 (125)	6.4 (31)	6.7 (37)	6.9 (30)	7.0 (29)	7.0 (25)	6.8 (16)
Maternal age at enrollment, % (n)
<20	5.1 (58)	4.4 (21)	4.5 (25)	3.2 (14)	3.1 (13)	3.3 (12)	2.1 (5)
20–25	32.1 (365)	29.4 (141)	29.6 (163)	29.9 (129)	29.3 (122)	33.5 (120)	33.8 (79)
26–30	38.2 (434)	40.1 (192)	40.2 (221)	43.1 (186)	42.3 (176)	38.5 (138)	40.2 (94)
>30	24.6 (280)	26.1 (125)	25.6 (141)	23.8 (103)	25.2 (105)	24.6 (88)	23.9 (56)
Maternal education at enrollment, % (n)
≤12 yr	31.3 (355)	28.2 (135)	28.2 (155)	24.6 (106)	25.0 (104)	25.1 (90)	25.8 (60)
>12 yr	68.7 (780)	71.8 (344)	71.8 (394)	75.4 (325)	75.0 (312)	74.9 (268)	74.3 (173)
Maternal smoking at enrollment, % (n)
Yes	17.2 (195)	15.1 (72)	14.2 (78)	13.7 (59)	13.9 (58)	15.1 (54)	13.7 (32)
No	82.8 (941)	84.9 (406)	85.8 (471)	86.3 (373)	86.1 (358)	84.9 (304)	86.3 (202)
Current smoking, % (n)
Yes	—	—	—	10.6 (43)	28.9 (120)	26.0 (92)	16.7 (39)
No	—	—	—	89.4 (362)	71.1 (296)	74.0 (262)	83.3 (195)

Definition of abbreviations: CC16 = club cell secretory protein; GM = geometric mean; GSD = geometric standard deviation; IQR = interquartile range; NO₂ = nitrogen dioxide.

Included participants had at least one CC16 measurement between age 6 and 32 years and their NO₂ exposure at enrollment or age 6 could be determined from their address. Participants were asked about their own smoking status beginning at age 16.

^*Number of participants with ≥1 CC16 measurement between age 6 and 32 years and NO₂ exposure was determined at enrollment address.

^†Number of participants with a CC16 measurement obtained during that follow-up period and their NO₂ exposure at enrollment or age 6 could be determined from their address.

An interquartile increase in NO₂ exposure at birth was associated with a decrease in CC16 levels from age 6 to 32 in both unadjusted and adjusted models (unadjusted, −4.8%; 95% confidence interval [CI], −8.6 to −0.7) (adjusted, −4.7%; 95% CI, −8.6 to −0.7) (Figure 1). Children in the highest quartile of exposure at birth had 8.6% lower (95% CI, −15.0 to −1.65) average levels of CC16 compared with children in the lowest quartile of exposure at birth. These trends were consistent in models stratified by follow-up period (see Figure E1). There was no significant association between children’s NO₂ exposures at age 6 and their CC16 levels from age 6 to 32 in either unadjusted or adjusted models (Figure 1). Results were consistent when models included only children with reported exposures for both time periods. When models were stratified by follow-up period, there was an increase in CC16 levels at age 6 with concurrent NO₂ exposures at age 6 (see Figure E1), and a decrease in CC16 levels at age 22–32. These trends were not significant.

Figure 1.

Percent change in club cell secretory protein levels from ages 6 to 32 years in association with an interquartile range increase in nitrogen dioxide exposure at birth and age 6 addresses, respectively. Associations are also shown per quartile. All models include random effects for subject and fixed effects for visit. Adjusted models include fixed effects for child’s race/ethnicity, child’s sex, and maternal smoking at enrollment. Participant counts include those with complete covariate, exposure, and outcome data. Percent change was calculated from the β coefficients according to methods for log-transformed outcome variables (24). CC16 = club cell secretory protein; CI = confidence interval; NO₂ = nitrogen dioxide.

We observed a significant interaction between race/ethnicity and NO₂ exposure at birth on CC16 levels (p for interaction = 0.04). The strongest NO₂/CC16 association was among black individuals, though this estimate was quite imprecise (−29.6%; 95% CI, −42.9% to −13.2%) (Figure 2). Our cohort had a relatively small black population (n participants = 26; n observations = 73), and thus these exploratory results should be interpreted with caution. Given that their exposures at birth were not significantly higher (see Table E2), this may indicate an increase in susceptibility to NO₂ exposure. Although we hypothesized that race may be a proxy for socioeconomic status in our cohort, we did not observe modification by maternal age, maternal education, or smoking. To ensure that this was not an effect of living in a neighborhood with lower socioeconomic status or poorer housing conditions, we also assessed effect modification by indices we previously derived from U.S. Census variables (27). There was no effect modification by these variables either. In our cohort, participants with any black parent were not more likely than other participants to have younger mothers, mothers who smoked, mothers with lower education, or live in neighborhoods with lower socioeconomic status or poorer housing. In a sensitivity analysis without black children, each interquartile increase in NO₂ exposure at birth was associated with a decrease in CC16 levels from age 6 to 32 years (adjusted, −3.5%; 95% CI, −7.5 to 0.7). Although not a significant association, these results are still suggestive of a negative association between NO₂ exposure and CC16 levels among children of other racial or ethnic heritage.

Figure 2.

Percent change in club cell secretory protein levels from ages 6 to 32 years in association with an interquartile range increase in nitrogen dioxide (NO₂) exposure at birth, by race/ethnicity. All models include random effects for subject, fixed effects for visit, child’s sex and maternal smoking at enrollment, and an interaction term for child’s race/ethnicity and NO₂ exposure at birth. Participant counts include those with complete covariate, exposure, and outcome data. Percent change was calculated from the β coefficients according to methods for log-transformed outcome variables (24). CC16 = club cell secretory protein; CI = confidence interval.

Discussion

To the best of our knowledge, this is the first study to prospectively document an association between early life NO₂ exposures and CC16 levels in humans. We demonstrated that ambient NO₂ exposures at birth address were associated with decreased circulating CC16 levels from age 6 to 32 years. However, there was no association between ambient NO₂ exposures at age 6 home address with CC16 levels. There was effect modification by race/ethnicity. Participants with at least one non-Hispanic black parent had significantly larger reductions in CC16 levels per interquartile range of NO₂ exposure compared with participants from other racial/ethnic backgrounds. CC16 may be a sensitive biomarker of early life damage to the respiratory tract following exposure to oxidizing air pollutants, such as NO₂.

Our study confirms findings from the only other study we identified that assessed the relationship between NO₂ exposure and CC16 levels in humans. Workers chronically exposed to nitrogen oxides at a nitric acid production factory were significantly more likely to have decreased levels of CC16, compared with unexposed workers (10). In our study, we demonstrated that increased NO₂ exposure levels at birth were associated with decreased CC16 levels through age 32 years. Of the participants included in our analyses with measured CC16 at enrollment and age 6, a total of 56% had moved residences between birth and age 6. However, exposures at age 6 were not significantly associated with CC16 levels from age 6 to 32 years. This supports conclusions from animal studies that there may be critical periods of lung development where epithelial cells may be uniquely susceptible to NO₂-induced death (1). Thus, children with increased NO₂ exposures during this critical window may be predisposed to having fewer club cells and reduced potential to produce CC16.

One of the complexities of CC16 as a biomarker of lung damage is the finding from in vitro and in vivo studies that show that acute or recent exposures to NO₂ may temporarily increase bronchial epithelial permeability, thereby allowing more CC16 into circulation, resulting in temporarily elevated levels (1, 9, 10). Although not significant, increased NO₂ exposures at age 6 were associated with increased CC16 levels at age 6 in our study (see Figure E1). However, similar to exposures at birth, increased NO₂ exposures at age 6 years were associated with decreased levels of CC16 in adulthood (age 22–32 yr). These findings indicate that the critical window of exposure may include more of early childhood.

Although there have not been any studies examining associations between ambient NO₂ exposure and CC16 levels in humans during critical periods of lung growth, a few studies in adults document that daily exposures to ambient particulate matter (PM) are associated with increased CC16 concentrations (28, 29). These associations are stronger with PM attributed to combustion sources (30), which are also more likely to be correlated with NO₂ in most ambient environments. Timonen and colleagues (29) and Jacquemin and colleagues (30) hypothesize that increased CC16 levels associated with ambient PM exposures on the same day may indicate that acute PM exposure is also associated with increased epithelial permeability. In one study, concurrent increase in CC16 levels with PM exposures are associated with temporary deficits in lung function (28), indicating that the immediate decrease in lung function following exposure to PM may be attributed to loss of pulmonary epithelium integrity (28). They demonstrated in a mediation model that increased levels of CC16 accounted for 4–36% of the lung function deficits associated with PM exposure. Thus, it is plausible that the positive association we observed in our cohort for concurrent NO₂ exposure and CC16 levels at age 6 may be explained by a similar mechanism (28), because NO₂ is often correlated with PM in ambient environments.

Previously, we reported that diminished CC16 levels (age 6–32 yr) in our cohort were associated with socioeconomic factors (i.e., Hispanic ethnicity, younger maternal age, maternal smoking, lower parental education attainment) (25). Although children with lower socioeconomic status are likely to have higher NO₂ exposures, in our study increased NO₂ exposure at birth was only associated with maternal smoking and none of the other factors. Interestingly, increased NO₂ exposure at age 6 was associated with all of these factors, yet there was not a significant association between NO₂ exposures at age 6 and longitudinal CC16 levels. In our cohort, socioeconomic status is not likely the explanation for the association of NO₂ exposure with CC16 levels (25). Future work is needed to determine if there are other factors that could explain these associations, such as additional environmental exposures, nutritional status, or access to health care.

In our analysis, children with any non-Hispanic black parent had significantly larger deficits in CC16 levels per interquartile increase in NO₂ exposures at birth compared with children from any other racial/ethnic background (see Figure E2). Because only 3.6% of our study population had any black parent, which is representative of Tucson, Arizona, further analyses in cohorts with a larger proportion of black children are needed to confirm this increased potential susceptibility.

However, there is a strong suggestion from the literature that black children may be more susceptible to NO₂ exposures. Multiple studies demonstrated significantly stronger associations between NO₂ exposures and hospital admissions or emergency room visits for asthma among black children compared with other children (31–34). This effect modification remained significant after controlling for insurance status and socioeconomic variables in a cohort of low-income children. Other studies demonstrated that racial disparities in childhood asthma are not just a function of income disparities (35–37). In a laboratory-controlled study, adult black males demonstrated significantly larger lung function deficits following exposure to ozone, another oxidizing air pollutant (38). Multiple hypotheses have been proposed, including differences in access to health care (39), oxidant defense genes (40, 41), or levels of micronutrients (42). For example, increased vitamin A levels are associated with increases in CC16 levels (43), and increased levels of vitamin E protect the lungs from the effects of NO₂ (1). Non-Hispanic black children in the United States have lower levels of vitamins A and E after controlling for income and socioeconomic factors, compared with children from other racial/ethnic backgrounds (44). Further work is needed to better understand why black individuals are more susceptible to oxidizing air pollutant (i.e., NO₂, ozone) exposures and the role that CC16 may have in this effect.

One of the key limitations of our study is that we only had exposure estimates for NO₂ and not for other air pollutants. Because air pollutants are often highly correlated, it is plausible that the effects we observed may be attributed to other air pollutants, such as ozone or PM. However, we were not able to assess for or control for those other exposures. Although there is biologic evidence of the effect of NO₂ on CC16 levels, club cells, and epithelial integrity, future studies should investigate exposures to mixtures of ambient air pollutants to determine their independent and joint effects on longitudinal CC16 levels.

In summary, in our nonselected birth cohort with 32 years of follow-up, we found that increased NO₂ exposure at birth but not at age 6 was associated with a decrease in CC16 levels through adult life (age 6–32 yr). Although the numbers were small, participants with at least one non-Hispanic black parent had significantly larger deficits in association with NO₂ exposure at their birth. Thus, CC16 may be a sensitive biomarker of early life lung damage from the oxidizing air pollutant NO₂ during critical windows of lung development, particularly among black children. Confirmation of these findings in other cohorts with a larger proportion of black participants could support the utility of CC16 as a means of identifying those children most susceptible to the effects of air pollution.