Exposures to Organophosphate Esters and Respiratory Morbidity among School-Aged Children with Asthma

Abstract

Organophosphate esters (OPEs) are an emerging class of chemicals used in a variety of consumer products as flame retardants, plasticizers, and additives. While prior epidemiologic studies suggest that OPEs may impact respiratory health, results remain inconclusive. We examined associations between urinary biomarkers of OPEs and symptoms of respiratory morbidity in a panel study of 147 predominantly Black school-aged children with asthma living in Baltimore City, Maryland. The study consisted of up to four seasonal, week-long, in-home visits where urine samples and self-reported asthma symptoms were collected on days 4 and 7 (n_samples = 438). We quantified concentrations of nine urinary OPE biomarkers: bis(2-chloroethyl) phosphate (BCEtp), bis(1-chloro-2-propyl) phosphate (BCPP), bis(1,3-dichloro-2-propyl) phosphate (BDCIPP), di-n-butyl phosphate (DBuP), di-benzyl phosphate (DBzP), di-o-cresylphosphate (DOCP), di-p-cresylphosphate (DPCP), di-(2-propylheptyl) phthalate (DPHP), and 2,3,4,5-tetrabromo benzoic acid (TBBA). We estimated prevalence odds ratios (POR) of respiratory morbidity symptoms using logistic regression with generalized estimating equations to account for our repeated measure design. We assessed BDCIPP and DPHP as continuous (log₂) concentrations and dichotomized exposure of BCEtP, DBuP, and DPCP (detect vs non-detect) based on their lower detection frequencies. We adjusted models for season, visit day, age, gender, caregiver education, health insurance type, exposure to household smoking, atopy, and PM_2.5. Higher DPHP concentrations were significantly associated with odds of daytime symptoms (POR: 1.26; 95% CI: 1.04–1.53; p = 0.02) where daytime symptoms consisted of trouble breathing due to asthma, reporting bother caused by asthma, and/or limitation in activities due to asthma. DBuP detection was associated with use of rescue medication on the day of sample collection (POR: 2.36; 95% CI: 1.05–5.29; p = 0.04). We also observed several consistent, albeit non-significant (p > 0.05), positive associations for BCEtP and DPCP and respiratory morbidity measures. This is the first study to evaluate the relationship between OPE biomarkers and respiratory morbidity symptoms in children with asthma, and findings suggest that further studies are warranted to confirm whether these associations are causal.

Graphical Abstract

1. INTRODUCTION

Organophosphate esters (OPEs) are an emerging class of high production volume chemicals first used in the early 2000s as replacements for polybrominated diphenyl ethers.^1–3 OPEs primarily serve as flame retardants, plasticizers, and additives in a wide range of consumer products (e.g., food packaging, furniture, clothing, floor polishes, electronics, children’s toys, personal care products, textiles, and building materials).^1–4 Since OPEs are not chemically bonded to consumer products, they can easily volatize into environmental media, including house dust,^5–8 where humans may be exposed primarily through inhalation and indirect ingestion.⁹ Their widespread use in consumer products has resulted in ubiquitous exposure in the U.S. general population,¹⁰ and recent biomonitoring studies indicate that, compared to adults, children may be disproportionately impacted by OPE exposures,^4,9,11–14 with Black children experiencing higher exposures compared to White children.¹² Greater exposures in children may result from increased hand-to-mouth behavior and greater inhalation and metabolic rates.^9,14,15 As such, these exposure patterns raise concerns for the impact of OPEs on pediatric respiratory health.

A limited number of epidemiological studies have reported associations between OPE exposures and the prevalence of respiratory outcomes, including asthma and allergies.^16–18 A cross-sectional study among adults and children reported significant associations between house dust exposures to tris(1-chloro-2-propyl) phosphate (TCPP) with atopic dermatitis and tributyl phosphate (TBP) with asthma and allergic rhinitis.¹⁶ Another cross-sectional study among school-aged children reported significant associations between house dust tris(1,3-dichloroisopropyl) phosphate (TDCIPP) with eczema and urinary bis(1,3-dichloro-2-propyl) phosphate (BDCIPP) (a primary biomarker of TDCIPP), the sum of TCPP biomarkers [including bis(1-chloro-2-propyl) phosphate (BCPP)], and hydroxy tris(2-butoxyethyl) phosphate (TBOEP-OH) with allergic symptoms including rhinoconjunctivitis and eczema.¹⁷ Finally, a follow-up study examining the sum of urinary biomarkers for TCPP and triphenyl phosphate (TPHP) reported significant positive associations with rhinoconjunctivitis.¹⁸ While several biological mechanisms may underlie these associations, the exact mechanism has not been clearly elucidated. Laboratory studies report that OPEs are involved in the development and exacerbation of asthma by eliciting increased allergic responses through immunomodulatory effects and oxidative stress, both of which are recognized risk factors in the pathophysiology of asthma.^19–22 A recent nested case-cohort study among children (ages 9–15 years) also reported significant positive associations between the OPE urinary biomarkers 2-ethylhexyl phenyl phosphate (EHPHP), bis(2-butoxyethyl) phosphate (BBOEP), and diphenyl phosphate (DPHP) with increased levels of oxidative stress biomarkers (i.e., 8-hydroxy-2′-deoxyguanosine, hexanoyl-lysine, and 4-hydroxynonenal).²³

While there is some evidence to suggest that exposures to environmental contaminants like OPEs may increase the prevalence of childhood respiratory symptoms and allergies, research is also needed to elucidate the role of environmental contaminants, including OPEs, in the risk of respiratory morbidity among children with pre-existing respiratory conditions who may be uniquely vulnerable to environmental exposures. In the present study, we sought to examine associations between OPE urinary biomarkers with symptoms of respiratory morbidity in a population of 147 predominantly Black school-aged children with persistent asthma, living in Baltimore City, Maryland.

2. METHODS

2.1. Study Design and Study Population.

We used data and banked biospecimens from children participating in the Domestic Indoor PM and Childhood Respiratory Morbidity (DISCOVER) study, which sought to examine the role of environmental pollutants in respiratory morbidity among children in Baltimore City, MD. The study has been described elsewhere.^24,25 Briefly, a total of 180 children, 150 children with asthma (100 atopic and 50 non-atopic) and 30 children without asthma, between the ages of 5 and 12 years were recruited from nine contiguous zip codes in Baltimore City between 2009 and 2013. For the present study, we limited our analysis to children with asthma (n = 150) to address our main research question of whether exposure to OPEs exacerbates respiratory symptoms among children with existing asthma. Inclusion criteria for children with asthma included a previous physician diagnosis of asthma with symptoms of asthma and/or reliever medication use during the six months preceding enrollment. Exclusion criteria included having a current diagnosis of another major pulmonary disease, plans to relocate residence during the study, consumption of antioxidant supplements, or inability to adhere to study protocols. The study was approved by the Johns Hopkins Institutional Review Board, and study staff obtained written informed consent from the parents or legal guardians of study participants prior to any data and sample collection.

As part of the study, participants completed a baseline clinic visit as well as week-long home visits occurring at baseline and every three months thereafter for a total of up to four, week-long home study visits (a schematic of the study design and framework is available in Supplemental Material Figure S1). Visits were designed to take place every three months to allow for assessment of seasonal variability for both environmental exposures and respiratory outcomes. Each week-long visit consisted of seven consecutive days (Sunday–Saturday). Study staff collected urine samples and respiratory symptom information on study participants on both days 4 (a weekday) and 7 (a weekend day), with each participant contributing up to eight visits/samples. During the baseline clinic visit, study staff also administered parental/guardian questionnaires to collect children’s demographic information including their age at enrollment, gender, race, health insurance type, body mass index (BMI), parental/guardian education level, and occurrence of smoking in the household. Allergic sensitization was also assessed at the baseline clinic visit by either allergic skin testing (ALK, Round Rock, Texas) or with serum allergen-specific IgE quantification corresponding to 14 aeroallergens (RAST, Pharmacia Diagnostics AB, Uppsala, Sweden). Atopy was defined as a positive sensitization to at least one individual allergen (IgE ≥ 0.35).²⁴ Because second-hand smoke exposure is an important risk factor for asthma exacerbations,²⁶ we accounted for both a parent-reported daily measure of smoke exposure (yes/no) and average weekly air nicotine concentrations, an objective measure of smoke in the home. We additionally included indoor PM2.5 (particulate matter with an aerodynamic diameter < 2.5 μm) concentrations in our models as this is a recognized risk factor for asthma exacerbations (Bose et al.²⁴). As described previously, (Bose et al.²⁴) indoor air measurements of particulate matter (PM) were collected at each home visit via gravimetric sampling to assess week-long average PM2.5, and air nicotine levels were measured with passive samplers and analyzed according to standard methods using gas chromatography with a nitrogen phosphate detector.

2.2. Respiratory Morbidity Outcome Measures.

Respiratory morbidity outcome measures were described in detail previously (Bose et al.²⁴). Briefly, we classified the child’s asthma severity as mild intermittent, mild persistent, moderate persistent, or severe persistent at the baseline visit based on the National Asthma Education and Prevention Program guidelines.^24,27 We also ascertained information on asthma medication use in the two weeks prior to baseline (e.g., including using at least one medication: albuterol, inhaled corticosteroid, or other medication). During each day of the week-long visit, parents/guardians completed a validated pediatric asthma diary twice daily to record the child’s daily respiratory symptoms. Study staff then reviewed the information provided in the diaries with parents/guardians on the following day. Daily respiratory symptoms included daytime symptoms (i.e., trouble breathing, whether the child was bothered by any asthma-related symptoms, and whether the child had to limit activities due to asthma-related symptoms), nighttime symptoms (being awoken by asthma symptoms), and use of albuterol and/or prednisone. Daytime symptoms were measured on a 6-point Likert scale and dichotomized as being absent or present, with composite daytime symptoms representing none versus any positive response across the three individual daily daytime symptoms (i.e., trouble breathing, whether the child was bothered by any asthma-related symptoms, and whether the child had to limit activities due to asthma-related symptoms). Nighttime symptoms included an assessment of the number of awakenings caused by asthma symptoms on a 4-point Likert scale, dichotomized for the present analyses as being absent or present. Study staff also captured daily information on healthcare utilization (i.e., any acute care, emergency department, or hospital visit for asthma).

2.3. OPE Exposure Assessment.

Spot urine samples were collected at random from all children in polypropylene urine collection cups on day 4 (weekday sample) and day 7 (weekend sample) and subsequently aliquoted into 2 mL cryovials and stored at −80 °C until shipment on dry ice to the analytical laboratory, NSF International (Ann Arbor, MI). The laboratory method has been described in detail elsewhere.²⁵ Briefly, concentrations (ng/mL) of nine urinary OPE biomarkers were quantified using liquid chromatography–tandem mass spectrometry (LC/MS/MS) including bis(2-chloroethyl) phosphate (BCEtp), BCPP, BDCIPP, di-benzyl phosphate (DBuP), di-benzyl phosphate (DBzP), di-o-cresylphosphate (DOCP), di-p-cresylphosphate (DPCP), DPHP, and 2,3,4,5-tetrabromo benzoic acid (TBBA). Standards of known purity and identity were used during the preparation of the calibration, quality control (QC), and internal standards. The laboratory followed QC procedures described by Kannan et al.²⁸ Briefly, QC pools were run in every batch, and during sample batch analysis, the matrix-specific QC samples (n = 35) were analyzed in duplicate at three levels (low-, mid-, and high-level concentrations) for each analyte. Low-level QC pool concentrations were 5 ng/mL for BCEtP and BCPP and 0.5 ng/mL for all other analytes, mid-level QC concentrations were 15 ng/mL for BCEtP and BCPP and 5 ng/mL for all other analytes, and high-level QC concentrations were 30 ng/mL for all analytes. Coefficients of variation (CV) for the QC replicates over all batches for the low-, mid-, and high-level QCs ranged from 6.0 to 15.4%, 2.7 to 12.1%, and 5.8 to 11.5%, respectively [e.g., the CV% range of 6.0–15.4% for the low-level QC represents the lowest and highest precision for all analytes at the corresponding analyte QC level]. The limits of detection (LOD) were established by replicate injections of low-concentration standards and ranged from 0.05 ng/mL (DBzP and TBBA) to 2 ng/mL (BCEtP and BCPP). For the present analysis, we used machine-read values for concentrations below the LOD and corrected OPE biomarker concentrations for specific gravity (SG) using the formula C_sg = C × [(1.024 – 1)/(SG – 1)], where C_sg is the SG-corrected OPE biomarker concentration (μg/L), C is the OPE biomarker concentration (μg/L), SG is the specific gravity for each observed urine sample, and 1.024 is the mean SG for the study population.²⁹

2.4. Statistical Analyses.

We calculated descriptive statistics to summarize participant demographics, clinical characteristics, PM_2.5, air nicotine, and OPE biomarker concentrations, and prevalence of asthma morbidity measures on the day of urine collection. For biomarkers with an overall detection frequency ≥ 70% (BDCIPP and DPHP), we log₂-transformed concentrations replacing machine-read 0 values with 0.001 before transformation to maximize the sample size. We dichotomized biomarkers with an overall detection frequency < 70% as not detected vs detected (BCEtP, DBuP, and DPCP) and excluded those biomarkers with an overall detection frequency < 10% (BCPP, DBzP, DOCP, and TBBA) from further analyses (see Supplemental Material Table S1 for complete OPE biomarker concentration distributions).

For the main analyses, we relied on asthma symptom data collected on the same days as the urine samples (i.e., day 4 and day 7 symptoms). This analytic approach was informed by the short biological half-lives of OPEs (i.e., <12 h), prior evidence of intraindividual variability in biomarker concentrations over the study period (intraclass correlation coefficients, ICCs = 0.18–0.33), weak to moderate correlations for OPE biomarker concentrations or detection within a week (r_Spearman = 0.40–0.50; r_tetrachoric = 0.31–0.63) in the study population,²⁵ and to optimize exposure timing to match the potential onset ofacute symptoms. For our analysis of respiratory symptoms and urinary OPE biomarker concentrations, we first assessed univariate associations for each pairwise combination of symptoms and OPE biomarker on the day of sample collection (Supplemental Material Table S2). We dichotomized symptoms collected on days 4 and 7 as follows: use of albuterol (yes/no), nighttime symptoms (yes/no), and composite daytime symptoms (yes/no). We did not assess healthcare utilization as an outcome based on the low number of affirmative responses. We then used generalized estimating equations (GEE) to examine associations using multivariable logistic regression, which allowed us to account for the non-independence of repeated observations within children by treating “child” as a clustering variable and an assumed exchangeable working correlation matrix. To further account for within-week and potential seasonal variations in exposures or outcomes, we adjusted all models for the day of visit (i.e., day 4 or day 7) and season. From the multivariable logistic regression models, we estimated prevalence odds ratios (PORs, 95% CI) for each symptom of respiratory morbidity (yes/no) per log₂ increase in continuous OPE biomarker concentrations, where a doubling in the biomarker concentration represents approximately a one quartile difference in each POR or for detection (vs non-detect) for less frequently detected OPE biomarkers. We included several covariates a priori in our multivariable analyses, including age at study enrollment, gender, primary caregiver’s highest level of education (<high school, high school diploma or equivalent, or some college), type of health insurance (private/self-pay or public), smoking in the household (yes/no), atopy (yes/no), PM_2.5 (μg/m³), and air nicotine concentrations (μg/m³).

To ensure the robustness of our results, we conducted sensitivity analyses by assessing individual daytime symptoms (Supplemental Material Table S3). We also ran separate models additionally adjusting our main effect models for child BMI (categorical) (Supplemental Material Table S4) and asthma severity (Supplemental Material Table S5), using the same modeling approach described above where asthma severity was modeled categorically as mild intermittent, mild persistent, moderate persistent, and severe persistent, and the mild intermittent category served as the referent group. BMI was modeled as a categorical variable derived from BMI percentiles as calculated according to the CDC guidelines.³⁰ BMI was considered as part of our sensitivity analyses rather than in our main models because obesity is recognized to be a major risk factor of asthma in children,³¹ but it could also be on the causal pathway as we previously found that OPEs may be linked to measures of adiposity.¹² We conducted additional sensitivity analyses to assess averaged weekly OPE biomarker concentrations or weekly detection for less frequently detected OPE biomarkers and the presence of any symptoms during the week at each visit. For BDCIPP and DPHP, the weekly averaged OPE exposures for each visit consisted of the mean of day 4 and day 7 biomarker concentrations (if missing one day, then the concentration available for the other day was treated as the weekly average for that visit). For less frequently detected OPE biomarkers (BCEtP, DBuP, and DPCP), we modeled exposure as a binary (yes/no) variable classified as “yes” for that week if biomarker concentrations were ≥LOD for either day 4 or 7 (Supplemental Material Table S6). Asthma symptoms were modeled as binary outcomes (yes/no) where any report of an asthma symptom during a given week was reported as “yes”. Models were adjusted for the same covariates as the primary multivariable analyses but without the day of visit.

As a secondary analysis, we examined gender differences by including multiplicative interaction terms (OPE biomarker × gender) in our main regression models (Supplemental Material Table S7). These analyses were not considered as part of our main models due to our limited sample size. For all adjusted multivariable models, we applied a statistical significance criterion of p < 0.05 for main effects and p < 0.10 for effect modification. All analyses were conducted in SAS version 9.4 (SAS Institute, Inc., Cary, NC).

3. RESULTS

3.1. Study Sample Characteristics.

After excluding children with missing home visit questionnaire information, our final study sample was comprised of 147 participants, with OPE concentrations measured on a total of 438 observation days. Among these 147 children, the median age at enrollment was 9.5 years, about half were male (54%), most were Black (93%), and nearly half were overweight or obese (48%) (Table 1). In addition, 87% of participant households reported having a form of public health insurance, with 13% having a private health insurance plan. About 46% of primary caregivers reported their highest level of educational attainment as “high school graduate”. About half of all children reported having mild asthma (52%), with 78% of all participants reported taking at least one asthma medication in the two weeks preceding the baseline clinic visit. Finally, 67% had atopy (i.e., sensitization to one or more allergens) (Table 1).

Table 1.

Descriptive Statistics of Baseline Characteristics in Children with Asthma Participating in the DISCOVER Study (N = 147)

children’s characteristics	N (%)
age at enrollment
5–7 years	36 (24.5)
8–10 years	63 (42.9)
11–13 years	48 (32.7)
gender
female	67 (45.6)
male	80 (54.4)
race
Black	137 (93.2)
other	10 (6.8)
body mass index (BMI)
underweight	10 (6.8)
normal	65 (44.2)
overweight/obese	70 (47.6)
missing	2 (1.4)
health insurance
private/self-pay	19 (12.9)
public	128 (87.1)
caregiver’s highest education obtained
<high school	41 (27.9)
high school graduate	67 (45.6)
≥ some college	38 (25.9)
missing	1 (0.7)
asthma severity
mild intermittent	47 (32.0)
mild persistent	29 (19.7)
moderate persistent	43 (29.3)
severe persistent	28 (19.1)
asthma medication use (two weeks prior to baseline)
use of at least one medication (yes)	114 (77.6)
albuterol (yes)	98 (66.7)
inhaled corticosteroid (yes)	65 (44.2)
other (yes)a	44 (29.9)
atopy
none	44 (29.9)
≥1 allergen	298 (66.7)
missing	5 (3.4)

^aIncludes cromolyn, leukotriene modifier, theophylline, or oral corticosteroids.

Among asthma symptoms assessed at each study visit day (i.e., day of urine sample collection), few participants reported taking a puff of albuterol (16%), having a non-scheduled physician visit for asthma (2%), being awoken by asthma (9%), having trouble breathing due to asthma (18%), being bothered due to asthma (18%), or experiencing limitation in activities due to asthma (15%) (Table 2). On days of sample collection, over one-third (37%) occurred among participants who reported smoking in the home by a household member on the day of sample collection. In addition, mean (SD) household concentrations of PM_2.5 and air nicotine were 36.2 μg/m³ (53.7) and 0.76 μg/m³ (1.65), respectively.

Table 2.

Descriptive Statistics of Time-Varying Characteristics among 147 Participants from 438 Study Visit Days

symptoms or characteristics		N	%
did you take any puffs of albuterol today?	no	367	83.8
	yes	71	16.2
did you visit a doctor, emergency room, or hospital for asthma (other than a scheduled visit to the doctor) or treated with prednisone during the previous 24 h?	no	410	93.6
	yes	8	1.8
	missing	20	4.6
nighttime symptoms (awoken by asthma)	no	379	86.5
	yes	38	8.7
	missing	21	4.8
daytime symptomsa	no	330	75.3
	yes	92	21.0
	missing	16	3.7
trouble breathing today	no	342	78.1
	yes	80	18.3
	missing	16	3.7
bothered by asthma today?	no	344	78.5
	yes	78	17.8
	missing	16	3.7
asthma limited activities today (activities include any sort of physical activity: running, playing, jumping, sports, bike-riding, gym, etc.)?	no	356	81.3
	yes	66	15.1
	missing	16	3.7
did anyone smoke in the home today?	no	278	63.5
	yes	160	36.5
	mean (SD)
PM2.5 (μg/m3)	36.2 (53.7)
air nicotine (μg/m3)	0.76 (1.65)

^aThe daily symptom variable is a composite variable that was generated based on responses to the following symptom variables: trouble breathing due to asthma, bother caused by asthma, and limitation in activity due to asthma, where an affirmative response of “yes” indicates that the participant experienced at least one of these symptoms, a “no” response indicates that the participant did not experience any of these symptoms, and a “missing” indicates that no information was provided for any of these daily daytime symptoms.

3.2. Sample Collection Characteristics and OPE Biomarker Concentrations.

Study participants contributed a total of 438 urine samples across all four seasonal visits (fall, spring, summer, and winter) (Table 3). Of the nine urinary OPE biomarkers examined, BDCIPP and DPHP exhibited the highest overall detection frequencies (98.2 and 99.5%, respectively). Geometric mean SG-corrected concentrations for BDCIPP and DPHP were 1.36 and 1.91 ng/mL, respectively. Other OPE biomarkers were either infrequently detected (<23% for BCEtP, DBuP, and DPCP) or undetected in children’s urine samples (BCPP, DBzP, DOCP, and TBBA). We also observed some seasonal variation in concentrations or detection frequency for some biomarkers, including BDCIPP, DPHP, and BCPP (Supplemental Material Table S1). For example, BDCIPP concentrations were generally greatest in the summer, compared to other seasons (Supplemental Material Table S1).

Table 3.

Distribution of SG-Corrected Urinary OPE Biomarker Concentrations (ng/mL) for the 147 Study Participants (n = 438 Samples)^a,b

biomarker	LOD	n ≥ LOD	DF (%)	GM	Min	p 25	p 50	p 75	Max
BCEtP	2.00	98	22.3	<LOD	<LOD	<LOD	<LOD	<LOD	70.8
BDCIPP	0.10	430	98.2	1.36	<LOD	0.77	1.38	2.80	95.1
DBuP	0.50	47	10.7	<LOD	<LOD	<LOD	<LOD	<LOD	8.09
DPCP	0.10	68	15.5	<LOD	<LOD	<LOD	<LOD	<LOD	2.31
DPHP	0.11	436	99.5	1.91	<LOD	0.93	1.62	3.20	314.5

^aConcentrations below the LOD are based on machine-read values. For geometric mean calculations, values of 0 were set to 0.001.

^bAbbreviations: LOD, limit of detection; DF, detection frequency; GM, geometric mean.

3.3. Multivariable Associations between OPE Biomarkers and Respiratory Outcomes.

We observed several significant and positive associations for detection of higher OPE biomarkers concentrations with asthma-related symptoms (Table 4). Specifically, a doubling in DPHP concentrations was associated with increased prevalence odds of daytime symptoms (POR: 1.26; 95% CI: 1.04, 1.53; p = 0.02); results were comparable when assessing individual daytime symptoms [trouble breathing due to asthma (POR: 1.20; 95% CI: 1.01, 1.43; p = 0.04); bother caused by asthma (POR: 1.32; 95% CI: 1.11, 1.57; p = 0.002); and limitation in activities due to asthma (POR: 1.35; 95% CI: 1.12, 1.63; p = 0.002)]. In addition, DBuP detection was significantly associated with the use of rescue medication on the day of sample collection (POR: 2.36; 95% CI: 1.05, 5.29; p = 0.04). While non-statistically significant (i.e., p > 0.05), we also observed positive associations between BCEtP detection and each respiratory symptom assessed as well as between DPCP and being awoken due to asthma and daytime symptoms (Table 4). Associations between BDCIPP and symptoms of respiratory morbidity were all null (Table 4). Unadjusted associations were generally in the same direction as adjusted associations (Supplemental Material Table S2). In sensitivity analyses, results for individual daytime symptoms were comparable to those obtained when modeling daytime symptoms as a composite variable (Supplemental Material Table S3). Results also remained robust when adjusting for BMI (Supplemental Material Table S4) and asthma severity (Supplemental Material Table S5), with some effect estimates becoming stronger. Moreover, no appreciable changes to our final results were observed when modeling symptoms of asthma severity by the visit week with averaged weekly OPE exposures (Supplemental Material Table S6).

Table 4.

PORs (95% Confidence Interval) of Multivariable Associations between Asthma Severity Symptoms and SG-Corrected Urinary OPE Biomarker Concentrations^a,b

symptom	N (%)	BDCIPPc	DPHPc	BCEtPd	DBuPd	DPCPd
did you take any puffs of albuterol today?	71 (16)	0.89 (0.77, 1.03)	1.17 (0.93, 1.46)	1.51 (0.67, 3.42)	2.36 (1.05, 5.29)	1.07 (0.47, 2.45)
missing, n = 0		p = 0.11	p = 0.18	p = 0.32	p = 0.04	p = 0.87
awoken by asthma	38 (9)	1.02 (0.76, 1.37)	1.25 (0.97, 1.60)	1.45 (0.55, 3.86)	2.21 (0.77, 6.31)	1.72 (0.71, 4.15)
missing, n = 20		p = 0.87	p = 0.08	p = 0.45	p = 0.14	p = 0.23
daytime symptomse	92 (21)	0.94 (0.81, 1.09)	1.26 (1.04, 1.53)	1.38 (0.62, 3.08)	1.68 (0.81, 3.48)	1.62 (0.86, 3.08)
missing, n = 16		p = 0.40	p = 0.02	p = 0.43	p = 0.17	p = 0.14

^aBoldface = p < 0.05

^bModels are based on GEE where urinary OPE biomarker samples were matched with symptoms collected on days 4 and 7, respectively, among 147 children covering 438 total visit days; all models were adjusted for seasons and days to account for the panel study design as well as age at enrollment, gender, caregiver’s education, insurance type, smoking in the household, atopy, air nicotine, and PM_2.5.

^cBDCIPP and DPHP were evaluated as continuous log₂-transformed measurements.

^dBCEtP, DBuP, and DPCP were evaluated as binary (detect vs non-detect) measurements.

^eThe daytime symptom variable is a composite variable that was generated based on responses to the following symptom variables: trouble breathing due to asthma, bother caused by asthma, and limitation in activity due to asthma, where an affirmative response (yes) indicates that the participant experienced at least one of these symptoms and a “no” indicates that the participant did not experience any of these symptoms.

In secondary analyses conducted to assess effect measure modification by gender (Supplemental Material Table S7), we found that DPHP in urine was associated with more albuterol use on the day of specimen collection among boys only (POR_boys: 1.43; 95% CI: 1.03, 1.99 vs POR_girls: 0.99; 95% CI: 0.77, 1.26; p_interaction = 0.05) and that detection of DPCP in urine was associated with more reports of daytime asthma symptoms among girls only (POR_boys: 0.39; 95% CI: 0.12, 1.32 vs POR_girls: 2.87; 95% CI: 1.33, 6.20; p_interaction = 0.01).

4. DISCUSSION

To our knowledge, this is the first study to examine associations between OPE biomarker concentrations and symptoms of respiratory morbidity among children with asthma. Based on the relatively short half-life of most OPEs examined, we hypothesized that daily OPE exposures would have an acute impact on asthma symptoms in our study population. In this predominantly Black pediatric cohort, we observed suggestive evidence of positive associations between exposure to DPHP and DBuP and several symptoms of respiratory morbidity. We also observed consistent positive associations, albeit not statistically significant, for BCEtP and DPCP and asthma morbidity.

As reported previously,²⁵ OPE urinary biomarker concentrations in our study population were comparable to those reported in previous biomonitoring studies among children, including children aged 6–11 years from the U.S. general population. For example, geometric mean biomarker concentrations among U.S. children versus children in our study for the most frequently detected OPEs were 2.25 vs 1.43 ng/mL for BDCIPP and 1.69 vs 1.84 ng/mL for DPHP, respectively. Similar to other studies, BDCIPP and DPHP were the most frequently detected OPEs.

While there is a lack of epidemiologic studies evaluating OPE exposures and child asthma-related symptoms, several in vivo studies using an asthma mouse model have found evidence that exposures to TPHP and TDCIPP (parent compounds for DPHP and BDCIPP, respectively) may alter immunomodulatory processes and induce oxidative stress, effects possibly leading to increased allergic responses involved in the development and exacerbation of asthma.^19–22 Respiratory health effects of the parent compounds of DBuP, BCEtP, and DPCP remain widely understudied, warranting further analysis.

Although BDCIPP was among the most prevalent OPE biomarkers detected, we found no associations with symptoms of respiratory morbidity. Previous epidemiological studies examining exposures to TDCIPP (parent compound of BDCIPP) and respiratory outcomes among school-aged children have reported associations with allergic symptoms, including rhinoconjunctivitis and eczema.^17,18 Additionally, a follow-up study among 7-year-old children found that TDCIPP levels in dust were associated with an increased risk of wheeze, among those without any filaggrin gene mutations, a major predisposing factor for eczema and asthma.³² Cytotoxic TDCIPP concentrations in vitro have also been found to induce expression of antioxidant enzymes as well as suppress immunomodulatory molecules.¹⁹ Notably, geometric mean BDCIPP concentrations (uncorrected) were considerably lower in our study compared to U.S. children aged 6–11 years participating in the 2013–2014 National Health and Nutrition Examination Survey (NHANES) cycle (1.36 vs 2.25 ng/mL, respectively).¹¹ The lack of associations with BDCIPP may be due to the lower concentrations noted in our study. For DPHP, which was also frequently detected, we found consistent positive associations with daytime symptoms. A prior study among children aged 9–15 years found that greater concentrations of DPHP were associated with greater concentrations of oxidative stress biomarkers, with oxidative stress being a common feature of obstructive airway diseases like asthma.²³ The magnitude of associations we observed between OPE biomarkers and asthma exacerbations was lower compared to studies assessing other environmental chemicals in similar pediatric populations.^33,34 Additional epidemiological evidence is needed to further investigate the potential effects of TDCIPP exposures on asthma morbidity.

When assessing effect measure modification by gender, we found that DPHP levels were associated with albuterol use on the day of specimen collection among boys only, while detection of DPCP was associated with increased odds of daytime symptoms among girls but not boys, with similar patterns for other outcomes. Gender differences are well documented in asthma research, with pre-adolescent boys having a higher prevalence of asthma compared to girls^35,36 possibly due to increased allergic inflammation and dysanapsis (disproportionate scaling of airway dimensions to the lung volume) in boys compared to girls.³⁷ However, beginning in puberty, asthma prevalence tends to be higher in girls than boys,³⁵ with estrogen and progesterone being key mediators in these differences given their involvement in the allergic and inflammatory responses underlying asthma.³⁵ A recent study of U.S. adolescents from NHANES (2013–2014) reported associations between greater urinary concentrations of DBuP and DPHP and lower total testosterone and estradiol as well as DBuP and lower sex hormone binding globulin (SHBG) in both genders.³⁸ Additionally, greater DPHP concentrations were associated with greater SHBG and total testosterone to estradiol ratio among females only. Conversely, among males, greater BDCIPP, DBuP, and DPHP concentrations were associated with less total testosterone and estradiol concentrations, while BDCIPP and DPHP were associated with greater SGHB concentrations.³⁸ We did not have enough power or information on pubertal status to further stratify our analyses by age or pubertal status based on gender; however, it is plausible that changes associated with puberty among girls could influence the relationship between DPCP and asthma morbidity. Given our findings and the potential influence of gender and pubertal status on asthma in childhood and adolescence, both gender and pubertal status should be considered and examined in future studies of OPE exposures and asthma.

Our study is not without limitations, including potential unmeasured confounding. Because we relied on biobanked specimens and data that were already collected, we were not able to address potential sources of OPEs to account for any differences in exposure in our analyses. We were also unable to collect information on pubertal status to further assess how sexual maturation could impact study findings based on gender differences. It is also plausible that our results may not be generalizable to other populations given that our study focused on a predominantly Black cohort of children, from a limited geographic region, and with current asthma. Future studies should ensure representation of children from diverse population ancestries, socioeconomic backgrounds, and geographic settings to confirm study findings. We also measured OPE biomarker concentrations in spot urine samples collected on the same day as participants reported asthma symptoms. While this timing was designed to optimize our ability to detect acute exacerbations related to recent exposures, we were unable to examine potential chronic effects of OPE exposures on asthma symptoms. Further mechanistic studies are needed to ascertain the exposure period of potential relevance. Finally, the detailed protocol followed by our participants resulted in a relatively small sample size that may have limited our statistical power to detect associations. Nonetheless, this is the first epidemiological analysis to provide information regarding the relationship between OPE exposures and respiratory morbidity and may therefore provide the basis for future studies examining the impact of OPEs among children and other similar populations.

Despite study limitations, our study has many notable strengths. Few epidemiological studies have examined associations between OPE exposures and respiratory outcomes. While prior studies have reported evidence for an increased risk of atopic dermatitis and prevalence of asthma and allergic symptoms (e.g., rhinoconjunctivitis and eczema) as a result of exposure to OPEs, these studies focused on populations with no reported pre-existing respiratory conditions.^{16–18,23,32} In addition, our panel study design and serial urine specimen collection allowed us to account for both seasonal and within-week intraindividual variability of OPE biomarker concentrations in our pediatric population. This further allowed for the reduction of potential exposure misclassification when compared to the use of single measurements of exposure, which is of particular importance when considering the short half-life of OPE biomarkers.⁷ Additionally, recall bias of symptom recordings by parents/guardians was minimized by the administration of a validated pediatric asthma diary during each day of the visit-week evaluation period. Diary information was also confirmed by study staff. Finally, we were able to control multiple important demographic confounders (e.g., age at enrollment, gender, BMI, health insurance type, etc.) as well as atopy and indoor PM_2.5 and air nicotine measurements, which have been found to increase the risk of respiratory symptoms among patients with asthma.³⁹

In summary, we found consistent associations between DPHP and DBuP with symptoms of respiratory morbidity as well as several positive associations, albeit not statistically significant, for BCEtP and DPCP. With the increased use of OPEs in consumer products^7,40–42 and emerging evidence suggesting that OPE exposures may affect human respiratory health,^{16–18,23,32} our findings support the need for additional epidemiological studies to elucidate their impact on children’s respiratory health.

Supplementary Material

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acs.est.2c05911.

DISCOVER study framework for sample collection, distribution of dilution-corrected urinary OPE biomarker concentrations for study participants by season and collection day, and model results from our sensitivity and secondary analyses (PDF)