Dietary Phthalates and Low-Grade Albuminuria in US Children and Adolescents

Summary

Background

Low-grade albuminuria is an indicator of endothelial dysfunction and is associated with an increased risk of cardiovascular disease. A graded level of exposure to bisphenol A was recently identified to be associated with increased risk of low-grade albuminuria in children and adults. Because bisphenol A and phthalates coexist as dietary contaminants, this study investigated whether exposure to phthalates is also associated with low-grade albuminuria.

Design, setting, participants, & measurements

Data were examined from 667 children who participated in the 2009–2010 National Health and Nutrition Examination Survey and who had results for urinary phthalate metabolites and albumin excretion. Urinary albumin and creatinine concentrations were measured in a first morning specimen using a solid-phase fluorescent immunoassay and a Roche/Hitachi Modular P Chemistry Analyzer with an enzymatic method, respectively. Phthalate metabolites were analyzed in a separate spot urine sample from each participant, using high-performance liquid chromatography and tandem mass spectroscopy.

Results

For each (roughly) 3-fold increase in metabolites of di-2-ethylhexylphthalate (a high molecular weight phthalate commonly found in foods), a 0.55 mg/g increase in albumin/creatinine ratio (ACR) was identified (P=0.02), whereas a 1.30-fold odds of a higher ACR quartile was also identified for each (roughly) 3-fold increase (P=0.02). Higher ACR was not identified in relationship to metabolites of lower molecular phthalates commonly found in lotions or shampoos, suggesting specificity.

Conclusions

Although reverse causation and unmeasured confounders represent alternative explanations, these findings, in conjunction with our earlier data on bisphenol A, indicate that a wide array of environmental toxins may adversely affect albuminuria and potentially increase the risk of cardiovascular disease. In view of the potential long-term health implications of ongoing exposure in this vulnerable subpopulation, our data support both further study and renewed regulatory efforts to limit exposure during childhood.

Introduction

The metabolic syndrome, composed of increased body weight, insulin resistance, hypertension, and hyperlipidemia, is increasingly common in childhood (1). In contrast, CKD is unusual in pediatric patients (2). Although the number of children requiring renal replacement therapy is still very low (at 15 per million population in 2008), there is evidence that the incidence of CKD is increasing (3). The epidemic of obesity and metabolic syndrome in children may be contributing to the rising frequency of ESRD in this age group (4). Oxidant stress has been identified as a common factor in both conditions and is one of the key drivers of disease progression in the full spectrum of CKD (5,6). Despite extensive knowledge of pathophysiologic mechanisms behind metabolic syndrome and CKD progression and the link between the two clinical entities, there is still great variability in the natural history and response to treatment in children with these conditions.

Environmental chemicals have been identified that are injurious to the kidney, including lead, cadmium, and mercury (7). Bisphenol A (BPA) and phthalates are of particular concern for nephrotoxicity because of increasing laboratory and epidemiologic data suggesting that they produce oxidative stress (8–10). Exposure to these chemicals is readily modified through diet (11–15). BPA is used to manufacture resin used to coat food and beverage containers (16), whereas phthalates are esters of phthalic acid that are used in a broad array of consumer products and food uses (17).

Laboratory studies have found that phthalate metabolites increase release of IL-6, a proinflammatory cytokine (18), and expression of integrin in neutrophils (10). Biomarkers of phthalate exposure have also been associated with increases in C-reactive protein and γ-glutamyltransferase (19), and the oxidative stress markers, malondialdehyde and 8-hydroxydeoxyguanosine (20,21).

Epidemiologic studies corroborate the notion that BPA and phthalates are associated with metabolic syndrome, endothelial dysfunction, and albuminuria in healthy individuals. A study of 3055 Chinese adults living in the Shanghai region identified urinary BPA as an independent determinant of the extent of low-grade albuminuria (albumin/creatinine ratio [ACR] <30 mg/g), with 23% increased odds for the highest quartile of albuminuria in the highest urinary concentration BPA quartile compared with the lowest quartile (22). We recently confirmed the association of urinary BPA with albuminuria among children and adolescents, aged 6–19 years, in data from the 2009–2010 National Health and Nutrition Examination Survey (NHANES) (23). Finally, we have reported a direct relationship between urinary phthalates and BP in the 2003–2008 NHANES (13).

Investigations into the effect of exposure to environmental chemicals in otherwise healthy children and adolescents might identify a modifiable factor that can be alleviated to improve outcomes in pediatric patients with CKD. In this report, we expand on our previous studies of BPA and examine whether exposure to phthalates is also associated with increased risk of developing low-grade albuminuria.

Materials and Methods

Data Source and Sample

NHANES is a biannual, multicomponent, nationally representative survey of the noninstitutionalized US population administered by the National Center for Health Statistics (NCHS) of the US Centers for Disease Control and Prevention (CDC). Data from the 2009–2010 questionnaire, laboratory, and physical examination components were used in the present analysis. Of the 2596 children aged 6–19 years who participated, our analytic sample comprised 710 nonpregnant participants with urinary phthalate measurements and first morning urine samples with creatinine measurements. The New York University School of Medicine Institutional Review Board exempted this project from review on the basis of its analysis of an already collected and deidentified dataset.

ACR Measurement

Beginning in 2009–2010, NHANES collected a first morning urine sample in a subsample of participants, which enabled a standardized assessment of albuminuria without the confounding effect of orthostatic proteinuria. Urine specimens were then processed, stored, and shipped to the University of Minnesota (Minneapolis, MN) for analysis of urinary albumin using a solid-phase fluorescent immunoassay (24). Measurement of urinary creatinine was performed on a Roche/Hitachi Modular P Chemistry Analyzer using an enzymatic method.

We calculated the ACR, and log-transformed the ACR because it had a skewed distribution. In analyses of ACR as a continuous variable, we excluded participants with macroalbuminuria (≥300 mg/g, n=5) or microalbuminuria (30–300 mg/g, n=38). In separate analyses, we modeled the probability of having either macroalbuminuria or microalbuminuria as a categorical dependent variable.

Measurement of Urinary Phthalates

Phthalate metabolites were measured in one spot urine sample from each participant, and analyzed using high-performance liquid chromatography and tandem mass spectroscopy. This testing was done on a different urine sample than the first morning specimen used for determination of ACR. More extensive methodological description is provided elsewhere (25). For phthalate concentrations below the level of detection, we substituted the limit of detection divided by the square root of 2, as routinely assigned by NHANES. To adjust analyses for urine concentration, we included urinary creatinine, as determined in the spot sample, as a covariate (26,27).

We grouped urinary biomarkers for exposure according to their use in product categories (see Table 1). Low molecular weight (LMW) phthalates (diethylphthalate, di-n-butylphthalate, and di-n-isobutylphthalate) are frequently added to shampoos, cosmetics, lotions, and other personal care products to preserve scent (17). High molecular weight (HMW) phthalates (di-2-ethylhexylphthalate, di-n-octylphthalate, and butylbenzylphthalate) are used to produce vinyl plastics for diverse settings ranging from flooring, clear food wrap, and intravenous tubing (28). Within the HMW category, di-2-ethylhexylphthalate (DEHP) is of particular interest because industrial processes to produce food frequently use plastic products containing DEHP (29). Di-isononylphthalate (DINP) is another HMW phthalate widely used in food contact materials, whereas di-isodecylphthalate (DIDP) is used in furnishings, cookware, medications, and many other products (25).

Table 1.

Widely produced phthalates, their common uses and metabolites

Phthalate Parent Compound	Metabolite	Potential Sources Of Exposure	Category
Diethyl phthalate	Monoethylphthalate (MEP)	Cosmetics, pharmaceuticals/herbal coating, insecticide	LMW
Di-n-butyl phthalate and di-isobutyl phthalate	Mono-n-butylphthalate (MBP)	Cosmetics, pharmaceuticals	LMW
	Mono-isobutylphthalate (MiBP)		LMW
Di-n-butyl phthalate	Mono-n-butylphthalate (MBP)	Medicines, cosmetics, cellulose acetate plastics, latex adhesives	LMW
Butyl benzyl phthalate	Monobenzyl phthalate (MBzP)	Vinyl flooring, adhesives, sealants, food packaging, furniture upholstery, vinyl tile, carpet tiles, artificial leather, certain adhesives	HMW
Di-n-octyl phthalate	Mono(3-carboxypropyl) phthalate (MCPP)	Soft plastics	HMW
Di-isodecylphthalate (DIDP)	Monocarboxyisononyl phthalate (MCNP)	Furnishings, cookware, medications	DIDP (also HMW)
Di-isononylphthalate (DINP)	Mono(3-carboxypropyl) phthalate (MCPP)	Food contact materials	DINP (also HMW)
	Monoisononyl phthalate (MNP)
	Monocarboxyisooctyl phthalate (MCOP)
Di-2-ethylhexyl phthalate (DEHP)	Mono(2-ethylhexyl) phthalate (MEHP)	Medical tubing, food packaging, toys, wall coverings, floor tiles, furniture upholstery, shower curtains, garden hoses, swimming pool liners, rainwear, automobile upholstery	DEHP (also HMW)
	Mono-(2-ethyl-5-carboxypentyl) phthalate (MECPP)		DEHP (also HMW)
	Mono-(2-ethyl-5-hydroxyhexyl) phthalate (MEHHP)		DEHP (also HMW)
	Mono-(2-ethyl-5-oxohexyl) phthalate (MEOHP)		DEHP (also HMW)

LMW, low molecular weight; HMW, high molecular weight.

We calculated molar sums for LMW and HMW metabolites and, within the HMW category, molar sums for DEHP, DIDP, and DINP metabolites. We expressed the LMW molar sum of monomethylphthalate (MMP), monoethylphthalate (MEP), mono-n-butylphthalate (MBP), and mono-isobutylphthalate (MiBP). We calculated the HMW molar sum of mono-(3-carboxypropyl) phthalate (MCPP), mono-(2-ethyl-5-carboxypentyl) phthalate (MECPP), mono-(2-ethyl-5-hydroxyhexyl) phthalate (MEHHP), mono-(2-ethyl-5-oxohexyl) phthalate (MEOHP), mono-(2-ethylhexyl) phthalate (MEHP), and monobenzylphthalate (MBzP). We calculated the molar sum of DEHP by computing and adding molarities of MEHP, MECPP, MEHHP, and MEOHP, and calculated the DINP metabolite concentrations by adding molarities of monocarboxyisooctyl phthalate (MCOP), MCPP, and monoisononyl phthalate (MNP). The molarity of the monocarboxyisononyl phthalate (MCNP) was calculated to represent the DIDP concentration, because it is the main DIDP metabolite (30). Our primary exposure variables were the log-transformed total molar concentrations of LMW, HMW, DEHP, DIDP, and DINP metabolites, although we present results of individual metabolites in secondary analyses.

Control for Potential Confounders

Our analyses included an array of confounders that might be correlated with elevated ACR and/or phthalate exposure. The first of these was hypertension. In NHANES, certified examiners assess systolic (first Korotkoff phase) and diastolic (fifth Korotkoff phase) BP three times in children aged 8–19 years after they sit quietly for 5 minutes. A fourth attempt may be made if one or more of the initial measurements is incomplete or interrupted. We followed the common practice of averaging systolic and diastolic BP measurements for purposes of generating continuous systolic and diastolic BP variables as well as categorical prehypertension variables (31). We calculated systolic/diastolic BP z scores from mixed-effects linear regression models derived using data from the 1999–2000 NHANES (32). We input height z scores derived from CDC norms, sex, and age to compute expected systolic/diastolic BPs (33). We calculated BP z scores from the measured BPs using the following formula: z = (x – µ)/σ, where x is the measured BP, µ is the expected BP, and σ is derived from the same NHANES data. We categorized BP outcomes into present/absent prehypertension (BP ≥90th percentile for age/height z score/sex).

Because exposure to tobacco smoke is a risk for metabolic syndrome in adolescence (34), we included serum cotinine in multivariable models. We categorized children and adolescents into low (<0.015 ng/ml, the limit of detection for the 2003–2006 NHANES), medium (<2 and ≥0.015 ng/ml) and high (≥2 ng/ml) categories (35,36).

Race/ethnicity was categorized into Mexican American, other Hispanic, non-Hispanic white, non-Hispanic black, and other in the NHANES. Caregiver education was categorized as follows: less than 9th grade, 9th–12th grade, high school or graduate equivalency diploma, some college, and college or greater. Poverty-income ratio was categorized into quartiles, within the sample for which urinary phthalates was measured. Age was categorized into two groups: 6–11 years and 12–19 years.

In the NHANES, trained health technicians assessed body measurements, following standardized measurement procedures (37). Body mass index (BMI) z scores were derived from the 2000 CDC sex-specific norms, and overweight and obese were categorized as BMI z scores ≥1.036 (85th percentile for age and sex) and ≥1.64 (95th percentile), respectively (33,38).

Cholesterol measurement was performed using a Roche Modular P chemistry analyzer after applying an enzymatic method. A cutpoint of 170 mg/dl was applied to dichotomize participants into high and low subgroups, following the American Academy of Pediatrics National Cholesterol Education Program guidelines (39).

Fasting insulin and glucose are available in a partially overlapping subsample of participants (n=142). The homeostasis model of assessment–insulin resistance value was calculated from fasting insulin and glucose (measured only in adolescents), and we also utilized the cutpoint of 4.39 (>2 SD above mean homeostasis model of assessment–insulin resistance for normal-weight adolescents with normal fasting glucose) identified by Lee et al. in the 1999–2002 NHANES) (40). Recognizing our modest capacity to control for insulin resistance in this analysis, we created a “missing” category for this and all potential confounders to maximize sample size in multivariable analysis. Serum cotinine data were missing in 26.5% of participants, cholesterol data were missing in 11.4%, and BP data were missing in 17.0%. Otherwise, <5% of values were missing for each confounding variable.

Statistical Analyses

The sample was described, applying sample weights according to the NCHS guidelines to produce national estimates, adjusting for the oversampling of racial/ethnic minorities and young children (41). Appropriately weighted univariable, bivariable, and multivariable analyses were conducted in a fashion that reflects the complex survey sampling design, using Stata 12.0 (StataCorp, College Station, TX) survey (‘svy’) commands.

Our analyses of the association modeled the relationship between log-transformed urinary phthalate and ACR specified two ways: (1) quartiles of ACR in the population of children with phthalate metabolites, and (2) log-transformed ACR. We performed univariate regression analyses of the association between ACR and the potential confounders named above, as well as regression analyses of the association between urinary phthalate and potential confounders, first adjusting only for creatinine. Multivariable regression analyses predicating ACR were adjusted for urinary creatinine, demographic and exposure characteristics (race/ethnicity, age, caregiver education, poverty/income ratio, sex, and serum cotinine), as well as BMI, elevated cholesterol, prehypertension, and insulin resistance categories. Linear regression was used to model ACR as a continuous variable, logistic regression was used to model micro/macroalbuminuria as an outcome, and ordered logistic regression was used to model ACR quartile as an outcome.

Addition of Urinary BPA to Models, and Interaction Testing

Recognizing that exposure to BPA, phthalates, and other environmental chemicals does not occur in isolation, we added quartiled and log-transformed urinary BPA to final multivariable models. We also tested models with interaction terms of log-transformed BPA with log-transformed phthalate metabolite concentrations in regression analyses, controlling only for creatinine to avoid overspecification.

Sensitivity Analyses

To ensure that our results were not an artifact of statistical weighting, we reprised our analyses in unweighted modeling, which is well known to enhance statistical efficiency while maintaining validity, especially when the analyses are intended to determine associations within a sample population rather than estimating prevalence of a condition or mean for a biologic outcome (such as BP or insulin) (42).

On the basis of the BMI category- and race/ethnicity-specific rates identified in the NHANES sample, we also utilized multiple imputation techniques (43) to randomly generate a replacement value for each missing data point for categorical insulin resistance data. We performed 20 imputations for the imputed variable following common practice (44), and we reprised the full multivariable model using the ‘mi estimate’ command in Stata 12.0 to assess whether this affected relationships of log-transformed or quartiled urinary phthalate metabolites with ACR outcomes.

Results

Table 2 presents population characteristics as well as differences in phthalate biomarkers and ACR by these characteristics. The median ACR was 6.06 mg/g (interquartile ratio [IQR], 4.3–9.3); the median LMW, HMW, and DEHP were 0.509 µM (IQR, 0.267–1.10), 0.401 µM (IQR, 0.205–0.721), and 0.201 µM (IQR, 0.099–0.356), respectively, LMW metabolites were higher in female participants than in male participants, as was ACR. Younger children had higher HMW, DEHP, DINP, and DIDP metabolites as well as higher ACR. DIDP was slightly higher in whites and other Hispanics and DINP was lower in whites, although racial/ethnic differences in ACR were not identified. Higher cotinine levels were associated with higher LMW, HMW, and DEHP metabolites, but not with ACR. Higher caregiver education was associated with higher DINP and DIDP metabolites, but was not associated with ACR. Overweight children had higher LMW metabolites and lower ACR. Analyses of phthalate metabolites against ACR identified HMW, DEHP, and DIDP metabolites to be higher among children in higher ACR quartiles. Conversely, ACR was higher with increasing HMW, DEHP, and DIDP quartiles.

Table 2.

Study population characteristics in 2009–2010 NHANES (N=667)^a

Characteristic	Total, n (%)b	Mean Urinary LMW Phthalate (µmol)c	Mean Urinary HMW Phthalate (µmol)c	Mean Urinary DEHP Phthalate (µmol)c	Mean Urinary DIDP Phthalate (µmol)c	Mean Urinary DINP Phthalate (µmol)c	Mean ACR (mg/g)
Sex
Male	355 (51.7)	0.33	0.41	0.21	0.009	0.07	5.00
Female	312 (48.3)	0.41d	0.39	0.19	0.009	0.10	6.00d
Age (yr)
6–11	336 (43.2)	0.51	0.48	0.22	0.013	0.09	7.60
12–19	331 (56.8)	0.43	0.31e	0.15d	0.008e	0.06f	5.81e
Race/ethnicity
Hispanic-Mexican American	180 (13.1)	0.55	0.39	0.20	0.009	0.06	6.21
Hispanic-other Hispanic	82 (6.5)	0.73	0.50	0.18	0.010e	0.08	6.77
Non-Hispanic white	226 (58.2)	0.40	0.50	0.18	0.010d	0.04f	6.83
Non-Hispanic black	128 (12.9)	0.63	0.40	0.16	0.009	0.07	5.67
Other	51 (9.4)	0.45	0.50	0.19	0.008	0.05	6.21
Poverty/income ratio
Quartile 1 (<0.78)	114 (12.4)	0.54	0.38	0.19	0.008	0.06	6.22
Quartile 2 (0.78–1.35)	159 (17.8)	0.51	0.43	0.21	0.009	0.07	7.70d
Quartile 3 (1.35–2.91)	164 (24.2)	0.50	0.37	0.17	0.009	0.08	6.61
Quartile 4 (at least 2.91)	165 (36.9)	0.40	0.37	0.18	0.009	0.08	6.63
Caregiver education
Less than 9th grade	75 (6.6)	0.61	0.31	0.18	0.007	0.04	6.71
9th–12th grade	115 (12.9)	0.49f	0.35	0.18	0.009f	0.05	6.55
High school or GED	145 (20.9)	0.53	0.38	0.17	0.009f	0.08d	6.02
Some college	185 (30.8)	0.44f	0.38	0.18	0.009d	0.08d	6.21
College or greater	128 (28.8)	0.40d	0.40	0.20	0.012d	0.08d	7.43
Serum cotinine (ng/ml)
<0.015	170 (23.1)	0.41	0.33	0.16	0.009	0.07	6.95
0.015–1.9	159 (54.8)	0.47f	0.40d	0.20f	0.010	0.07	6.54
At least 2.0	161 (10.4)	0.58d	0.34	0.15	0.008	0.07	5.94
Obesity status
Not overweight	393 (63.3)	0.43	0.37	0.18	0.010	0.07	7.05
Overweight	267 (36.7)	0.59d	0.43	0.24	0.010	0.09	5.72f
BP
<90th percentile	521 (94.9)	0.44	0.35	0.17	0.009	0.07	6.75
≥90th percentile	32 (5.1)	0.57f	0.47	0.24	0.008	0.09	6.19
Cholesterol
Not elevated	390 (65.2)	0.47	0.38	0.19	0.009	0.07	6.84
Elevated	201 (34.8)	0.45	0.37	0.17	0.009	0.08	6.12
Insulin resistance
Not resistant	101 (72.3)	0.42	0.30	0.14	0.007	0.06	5.69
Resistant	40 (27.7)	0.55	0.50f	0.23	0.009	0.11f	5.25
ACR (mg/g)
Quartile 1 (<4.3)	170 (25.6)	0.45	0.30	0.15	0.008	0.06
Quartile 2 (4.3–6.06)	159 (23.4)	0.42	0.28	0.16	0.009f	0.07
Quartile 3 (6.06–9.34)	161 (22.8)	0.50	0.33f	0.21f	0.010f	0.08
Quartile 4 (≥9.34)	177 (28.1)	0.48	0.32f	0.22f	0.012f	0.08
LMW metabolite (µmol)c
Quartile 1 (<0.266)	179 (30.2)						6.43
Quartile 2 (0.266–0.509)	168 (26.6)						6.39
Quartile 3 (0.509–1.10)	162 (22.8)						6.80
Quartile 4 (≥1.10)	158 (20.5)						6.79
HMW metabolite (µmol)c
Quartile 1 (<0.205)	173 (26.5)						5.90
Quartile 2 (0.205–0.401)	166 (25.2)						5.92
Quartile 3 (0.401–0.721)	165 (24.9)						7.07f
Quartile 4 (≥0.721)	163 (23.4)						7.72f
DEHP metabolite (µmol)c
Quartile 1 (<0.100)	175 (26.8)						5.72
Quartile 2 (0.100–0.201)	164 (26.2)						6.31
Quartile 3 (0.201–0.356)	164 (24.7)						6.83
Quartile 4 (≥0.356)	164 (22.3)						7.82d
DIDP metabolite (µmol)c
Quartile 1 (<0.005)	167 (23.5)						5.54
Quartile 2 (0.005–0.010)	169 (28.0)						6.81f
Quartile 3 (0.010–0.019)	163 (25.8)						6.61
Quartile 4 (≥0.019)	168 (22.8)						7.47f
DINP metabolite (µmol)b
Quartile 1 (<0.031)	168 (22.7)						6.10
Quartile 2 (0.031–0.065)	174 (26.0)						6.18
Quartile 3 (0.065–0.145)	160 (24.5)						6.87
Quartile 4 (≥0.145)	165 (26.7)						7.16

LMW, low molecular weight; HMW, high molecular weight; DEHP, di-2-ethylhexyl phthalate; DIDP, di-isodecylphthalate; DINP, di-isononylphthalate; ACR, albumin/creatinine ratio; GED, general educational development.

^aThe total number of participants from some variables do not total to 667 due to missing data. See text.

^bAll percentages are weighted using population weights for the sample in which phthalate metabolites were measured.

^cCalculated using linear regression, after adjustment for urinary creatinine.

^dP<0.01.

^eP<0.001.

^fP<0.05.

Multivariable analyses (Table 3) confirmed the association of DEHP, but not other phthalate metabolite groupings, with increases in continuous or quartiled ACR measures. The fourth quartile of DEHP metabolite was associated with a 1.30 mg/g increase (P=0.01) as well as a 2.07-fold odds (95% CI, 1.15 to 3.72) of increased ACR quartile (P=0.02). Odds of micro/macroalbuminuria were significantly lower in the second DINP quartile (odds ratio, 0.36 [95% CI, 0.16 to 0.79]; P=0.02), although children and adolescents in the other DINP quartiles did not have significantly different ACR. When measured continuously, both HMW and DEHP metabolites were associated with increases in ACR. A 0.52 mg/g increase in ACR (P=0.04) was identified for each log unit (roughly 3-fold) increase in HMW metabolites, whereas a 0.55 mg/g increase in ACR was identified (P=0.02) for each log unit increase in the DEHP subcategory of HMW metabolites. A 1.30-fold odds (95% CI, 1.04 to 1.61) of higher ACR quartile was also identified per log unit increase in DEHP metabolites (P=0.02). No significant differences in micro/macroalbuminuria were identified.

Table 3.

Linear and logistic regression analysis of ACR outcomes against quartiled and log-transformed urinary phthalates

Phthalate Exposure Variable	Increment in ACR	Odds of Micro/Macroalbuminuria (95% Confidence Interval)	Odds of Higher ACR Quartile (95% Confidence Interval)
LMW quartile
1	Reference	Reference	Reference
2	−0.33 (−1.01 to +0.44)	3.08 (0.70 to 13.50)	0.85 (0.56 to 1.29)
3	+0.11 (−1.11 to +1.64)	3.29 (0.88 to 12.30)	1.21 (0.62 to 2.35)
4	+0.09 (−1.12 to +1.63)	2.56 (0.76 to 8.70)	0.99 (0.48 to 2.00)
HMW quartile
1	Reference	Reference	Reference
2	−0.23 (−1.22 to +0.98)	1.50 (0.26 to 8.51)	0.96 (0.47 to 1.93)
3	+0.69 (−0.36 to +1.94)	1.43 (0.43 to 4.76)	1.54 (0.79 to 3.00)
4	+1.02 (−0.18 to +2.48)	1.84 (0.30 to 11.48)	1.79 (0.87 to 3.65)
DEHP quartile
1	Reference	Reference	Reference
2	+0.25 (−0.49 to +1.09)	1.65 (0.32 to 8.63)	1.23 (0.73 to 2.06)
3	+0.31 (−0.70 to +1.56)	1.99 (0.35 to 11.20)	1.15 (0.54 to 2.45)
4	+1.30 (+0.30 to +2.47)a	1.67 (0.52 to 5.34)	2.07 (1.15 to 3.72)a
DIDP quartile
1	Reference	Reference	Reference
2	+0.91 (−0.09 to +2.09)	0.40 (0.08 to 2.00)	1.61 (0.89 to 2.93)
3	−0.70 (−0.75 to +2.61)	0.28 (0.06 to 1.35)	1.87 (0.87 to 2.01)
4	+0.97 (−0.37 to +2.67)	0.65 (0.16 to 2.69)	2.21 (0.91 to 5.36)
DINP quartile
1	Reference	Reference	Reference
2	−0.22 (−1.12 to +0.86)	0.36 (0.16 to 0.79)a	0.90 (0.46 to 1.78)
3	+0.11 (−1.02 to +1.51)	0.48 (0.09 to 0.38)	1.13 (0.59 to 2.20)
4	+0.56 (−0.49 to +1.81)	0.55 (0.16 to 0.31)	1.36 (0.74 to 2.51)
Increment per log unit
LMW	+0.16 (−0.29 to +0.65)	1.16 (0.75 to 1.80)	1.06 (0.84 to 1.35)
HMW	+0.52 (+0.02 to +1.05)a	1.18 (0.73 to 1.91)	1.29 (0.97 to 1.70)
DEHP	+0.55 (+0.11 to +1.010)a	1.11 (0.78 to 1.57)	1.30 (1.04 to 1.61)a
DIDP	+0.24 (−0.33 to +0.86)	0.85 (0.43 to 1.67)	1.20 (0.89 to 1.62)
DINP	+0.18 (−0.18 to +0.56)	0.99 (0.66 to 1.49)	1.11 (0.90 to 1.38)

All models control for sex, poverty/income ratio, caregiver education, serum cotinine, urinary creatinine, age, prehypertension, insulin resistance, body mass index, hypercholesterolemia, and race/ethnicity categories. ACR, albumin/creatinine ratio; LMW, low molecular weight; HMW, high molecular weight; DEHP, di-2-ethylhexyl phthalate; DIDP, di-isodecylphthalate; DINP, di-isononylphthalate.

^aP<0.05.

The examination of individual phthalate metabolites identified associations of three of the main DEHP metabolites with increases in continuous ACR and increased ACR quartiles in ordered logistic regression (Table 4). Roughly 3-fold increases in MEHHP, MEOHP, and MECPP were associated with 0.33 mg/g (P=0.02), 0.38 mg/g (P=0.01), and 0.38 mg/g (P=0.01) increases in ACR, and 24% (P=0.04), 28% (P=0.02), and 32% (P=0.02) increased odds of a higher ACR quartile. One LMW metabolite (MiBP, P=0.04) was significantly associated with increases in ACR, as was one non-DEHP HMW metabolite (MBzP, P=0.04). DIDP and DINP metabolites were not associated with increases in continuous ACR, and the only significant associations with increases in ACR quartile were identified in association with individual DEHP metabolites.

Table 4.

Linear and logistic regression analysis of ACR outcomes against log-transformed urinary phthalate metabolites

Increment per Log Unit	ACR Ratio	Odds of Higher ACR Quartile (95% Confidence Interval)
LMW metabolites
MMP	+0.19 (-0.06 to +0.45)	1.10 (0.94 to 1.28)
MEP	+0.04 (-0.26 to +0.36)	1.01 (0.85 to 1.21)
MBP	+0.28 (-0.10 to +0.69)	1.17 (0.90 to 1.52)
MiBP	+0.43 (+0.01 to +0.88)a	1.23 (0.97 to 1.57)
DEHP metabolites
MEHP	+0.23 (-0.14 to +0.2)	1.14 (0.92 to 1.42)
MEHHP	+0.33 (+0.05 to +0.62)a	1.24 (1.01 to 1.51)a
MEOHP	+0.38 (+0.10 to +0.68)a	1.28 (1.05 to 1.54)a
MECPP	+0.38 (+0.10 to +0.67)a	1.32 (1.06 to 1.65)a
DIDP metabolites
MCNP	+0.20 (-0.27 to +0.70)	1.20 (0.89 to 1.62)
DINP metabolites
MCPP	+0.24 (-0.13 to +0.64)	1.17 (0.91 to 1.52)
MiNP	+0.17 (-0.05 to +0.40)	1.09 (0.93 to 1.27)
MCOP	+0.10 (-0.19 to +0.40)	1.08 (0.89 to 1.30)
Other HMW metabolites
MBzP	+0.23 (+0.002 to +0.65)a	1.12 (0.97 to 1.29)
MCHP	+0.48 (-0.08 to +1.09)	1.40 (0.94 to 2.09)
MOP	+0.92 (-0.21 to +2.27)	3.06 (0.86 to 10.84)

All models control for sex, poverty/income ratio, caregiver education, serum cotinine, urinary creatinine, age, prehypertension, insulin resistance, body mass index, hypercholesterolemia, and race/ethnicity categories. ACR, albumin/creatinine ratio; LMW, low molecular weight; MMP, monomethylphthalate; MEP, monoethylphthalate; MBP, mono-n-butylphthalate; MiBP, mono-isobutylphthalate; DEHP, di-2-ethylhexyl phthalate; MEHP, mono(2-ethylhexyl) phthalate; MEHHP, mono-(2-ethyl-5-hydroxyhexyl) phthalate; MEOHP, mono-(2-ethyl-5-oxohexyl) phthalate; MECPP, mono-(2-ethyl-5-carboxypentyl) phthalate; DIDP, di-isodecylphthalate; MCNP, monocarboxyisononyl phthalate; DINP, di-isononylphthalate; MCPP, mono(3-carboxypropyl) phthalate; MiNP, monoisononyl phthalate; MCOP, monocarboxyisooctyl phthalate; HMW, high molecular weight; MBzP, monobenzyl phthalate; MCHP, mono-cyclohexyl phthalate; MOP, mono-n-octyl phthalate.

^aP<0.05.

Associations of DEHP with ACR were robust to removal of survey weights and multiple imputation of insulin resistance (Supplemental Material). Adjustment for log-transformed BPA attenuated the magnitude of the association (−12%) with continuous ACR but it remained significant (P=0.03). Adjustment for BPA in quartiles also attenuated the association of the fourth quartile of DEHP (−13%) with continuous ACR but the association remained significant (P=0.02). The addition of quartiled BPA in ordered logistic regression in full multivariable models failed to diminish significant association of the fourth quartile DEHP with increased ACR quartile (P=0.03), and log-transformed DEHP was significantly associated with higher ACR quartile (P=0.02) despite the addition of log-transformed BPA.

In each of these four models, quartiled/log-transformed BPA was positively but not significantly associated with ACR (P≥0.23). In particular, the addition of log-transformed DEHP attenuated the magnitude of the association of BPA with continuous ACR by 40% (P=0.23), whereas the addition of quartiled DEHP attenuated the association of the fourth quartile of BPA with continuous ACR by 33% (P=0.48). Although this result may reflect modest power for simultaneous study of BPA and DEHP, an alternative explanation is that BPA exposure may be a marker for other phthalate and other toxic exposures.

When an interaction term was added to models with log-transformed BPA and DEHP, controlled for urinary creatinine, DEHP remained significantly associated with increased ACR (P=0.01), BPA was associated with increased ACR at P=0.07, and the interaction term was not significant (P=0.40). Ordered logistic regression of the two continuous measures of BPA and DEHP with an interaction term yielded a similar pattern.

Discussion

In this report, we documented that graded exposure to phthalates is associated with increased low-grade albuminuria in healthy children and adolescents who participated in the 2009–2010 NHANES. The identified associations suggest that they are specific to metabolites of DEHP that are commonly found as food contaminants (45), and parallel our recent observation of increased low-grade albuminuria in relation to the degree of exposure to BPA (another dietary contaminant) in the same population (23). The effect size was similar for DEHP metabolites and BPA.

This study suggests that exposures to BPA and phthalates may represent modifiable risk factors that can affect the incidence and progression of conditions such as the metabolic syndrome and CKD. Response to treatment for, based on sustained changes in kidney function and urinary protein excretion, has failed to show substantial improvement despite extensive clinical research and the introduction of new treatments such as tacrolimus, mycophenolate mofetil, and rituximab (46,47). This underscores the importance of increased attention to these ubiquitous molecules because of the potential long-term ramifications of environmental exposure, especially in conjunction with other risk factors such as obesity or hypertension that comprise the metabolic syndrome.

The limitations of cross-sectional analyses cannot be underestimated. Alternative explanations of our results not only include unmeasured confounding, but also reverse causation (children with higher ACRs may simply excrete more phthalate metabolites). Oxidative stress biomarkers such as F2-isoprostane and 8-hydroxydeoxyguanosine are unavailable for the 2009–2010 NHANES for corroboration of potential mechanisms. The sample size is modest and limits the ability to evaluate the effect of phthalates in patient subgroups. The t_1/2 of phthalates is short (typically 12–48 hours) and a single urinary marker as studied here may not reflect the magnitude of low-grade albuminuria, a condition that accumulates over time. Urinary creatinine was included in models to adjust for urine concentration, but we also recognize that creatinine excretion also varies according to age (especially in growing children), muscle mass, and race/ethnicity. Specific gravity is unfortunately not available in the urine samples from the 2009–2010 NHANES (48).

A cross-sectional study cannot identify the mechanism of action by which exposure to phthalates or BPA may increase the risk of low-grade albuminuria. We suggest that a generalized increase in oxidant stress and subsequent endothelial function may account for the adverse effects of phthalates and BPA. Nonsignificance of interaction terms could result from separate mechanisms with additive effects or alternatively from a common mechanism with additive effects. This question cannot be resolved by our study design, and the failure to detect a significant interaction could also have been the byproduct of insufficient statistical power. Because low-grade albuminuria is a significant risk factor for the development of cardiovascular and renal disease (49–51), this underscores the importance of investigating the renal consequences posed by environmental chemical exposures. The benefits of preventing such exposures in children may be much greater given that they are more susceptible to the consequences of life-long exposure to environmental chemicals. Studies of genes involved in the regulation of inflammation and oxidant stress may help to identify patients at greatest risk of exposure to environmental toxins like phthalates and BPA.

Recently, using cross sectional data from 766 fasting 12–19 year olds in the 2003–2008 NHANES survey, Trasande et al. (52) demonstrated that, after controlling for demographic and behavioral factors, diet, age, BMI category, and urinary cotinine, for each log increase in urinary excretion of DEHP metabolites there was a 0.27 increase in HOMA-IR (P<0.001). Compared with the lowest tertile of DEHP excretion, the third tertile had a 49% increase in the prevalence of insulin resistance (P<0.025). These observations supplement our findings ad highlight the association of urinary phthalate excretion and cardiovascular risk factors in pediatric patients.

In conclusion, we have demonstrated that graded exposure to phthalates, like BPA, is associated with increased risk of low-grade albuminuria in healthy children and adolescents. This finding expands the roster of molecules that may have an adverse effect on the kidney. Given the recent association of DEHP metabolites with increased BP (13), this study adds to concerns about the adverse cardiovascular effects of phthalates. Because the presence of low-grade albuminuria and hypertension are risk factors for cardiovascular disease and CKD, these observations support the possible need for additional regulatory attention to environmental chemicals like phthalates and BPA.

Disclosures

H.T. is a consultant to Retrophin Inc. and Kaneka Corp.