Association of Exposure to Di-2-Ethylhexylphthalate Replacements With Increased Insulin Resistance in Adolescents From NHANES 2009–2012

Abstract

Context:

Di-isononyl phthalate (DINP) and di-isodecyl phthalate (DIDP) are environmental chemicals increasingly used to replace di-2-ethylhexylphthalate (DEHP) and commonly found in processed foods. Phthalate exposures, in particular DEHP, have been associated with insulin resistance in adolescents, but there are no data regarding the two substitutes, DINP and DIDP.

Objective:

This study aimed to examine associations of DINP, DIDP, and DEHP with insulin resistance outcomes.

Design, Setting, and Participants:

This was a cross-sectional analysis of 2009–2012 National Health and Nutrition Examination Surveys (NHANES) composed of 356 fasting 12–19-year-olds.

Main Outcome Measures:

Insulin resistance as a categorical outcome expressed as homeostatic model assessment of insulin resistance (HOMA-IR), using a cut point of 4.39 to define insulin resistance. We also examined continuous HOMA-IR as an outcome in secondary analyses.

Results:

Controlling for demographic and behavioral factors, diet, age, body mass index, and urinary creatinine, for each log increase in DINP metabolite, a 0.08 (P = .001) increase in HOMA-IR was identified. Compared with the first tertile of DINP (23.4% adjusted prevalence), the third tertile was associated with a 34.4% prevalence (95% confidence interval [CI], 27.3–41.6%; P = .033) of insulin resistance. Similarly, compared with the first tertile of DEHP (20.5% adjusted prevalence), the third tertile had 37.7% prevalence (95% CI 29.8–45.6%; P = .003).

Conclusions:

Urinary DINP concentrations were associated with increased insulin resistance in this cross-sectional study of adolescents. The previously identified association of DEHP with insulin resistance was also confirmed. Further, longitudinal studies are needed to confirm these associations, with the possibility to assess opportunities for intervention.

Phthalates are chemical compounds with a diverse array of uses and found in variety of consumer products, such as plastics, cosmetics, cleaning products, and building materials (1, 2). Routes of human exposure vary (oral, dermal, inhalation) depending on the specific compound, and phthalates can be classified into low-molecular weight (LMW) phthalates, which are frequently added to preserve scent (3), and high-molecular weight (HMW) phthalates, which are used to produce vinyl plastics for diverse applications ranging from flooring, clear food wrap, and intravenous tubing (Table 1) (4). Within the HMW phthalate category, di-2-ethylhexylphthalate (DEHP) is of particular interest because industrial processes to package food frequently use plastic products containing DEHP (5), and dietary intake from contaminated food is the largest contributor to exposure in children (4, 6).

Table 1.

Widely Produced Phthalates, Main Metabolites, and Common Uses

Phthalate Parent Compound	Metabolite	Potential Sources of Exposure	Category
Diethyl phthalate	Monoethylphthalate (MEP)	Cosmetics, nail polish, deodorant, perfumes/cologne, lotions, aftershave, pharmaceuticals/herbal coating, insecticide	LMW
Di-n-butyl phthalate and di-isobutyl phthalate	Mono-n-butylphthalate (MBP)	Nail polish, makeup, aftershave, perfumes, pharmaceuticals/herbal coating, chemiluminescent glow sticks	LMW
	Mono-isobutylphthalate (MiBP)		LMW
Di-n-butyl phthalate	Mono-n-butylphthalate (MBP)	Medicines, cosmetics, cellulose acetate plastics, latex adhesives, in nail polish and other cosmetic products, as a plasticizer in cellulose plastics, as a solvent for certain dyes	LMW
Butyl benzyl phthalate	Monobenzyl phthalate (MBzP)	Vinyl flooring, adhesives, sealants, food packaging, furniture upholstery, vinyl tile, carpet tiles, and artificial leather and is also used in certain adhesives	HMW
Di-n-octyl phthalate	Mono(3-carboxypropyl) phthalate (MCPP)	Soft plastics	HMW
DEHP	Mono(2-ethylhexyl) phthalate (MEHP)	PVC-containing medical tubing, blood storage bags, medical devices, food packaging, plastic toys, wall coverings, tablecloths, floor tiles, furniture upholstery, shower curtains, garden hoses, swimming pool liners, rainwear, baby pants, dolls, some toys, shoes, automobile upholstery and tops, packaging film and sheets	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)
DINP	Monoisononyl phthalate (MNP)	Plasticizer in polyvinyl chloride plastics. Paints, lacquers, and adhesives	HMW
	Monocarboxyisooctyl phthalate (MCOP)
DIDP	Monocarboxyisononyl phthalate (MCNP)	Plasticizer in polyvinyl chloride plastics. Paints, lacquers, and adhesives	HMW

Abbreviation: PVC, polyvinyl chloride.

Adapted from Braun et al, 2013 (65), and Trasande et al, 2013 (28).

Given that evidence and concerns about potential health risks related to DEHP exposure are increasing (7), DEHP is being replaced with alternative compounds, such as di-isononyl phthalate (DINP) and di-isodecyl phthalate (DIDP) (8). DINP and DIDP are also classified as HMW phthalates and are mainly used as polyvinyl chloride plasticizers; available estimates indicate that these compounds account for 33% and 63% of the plasticizer market in the United States and the European Union, respectively (8, 9). The progressive replacement of DEHP with DINP and DIDP is reflected in biomonitoring data showing a decrease in the levels of DEHP metabolites by 17–37% between 2001 and 2010 in national survey samples (10).

Early life exposure to phthalates has been associated with a variety of adverse effects, particularly involving endocrine processes (3, 11). Increasing evidence also points to a possible adverse effect of phthalates on glucose homeostasis and insulin resistance (12), in addition to a possible contributing role in the development of obesity, as shown by recent data reporting an association between urinary levels of phthalates and higher odds for obesity (body mass index [BMI]) in children and adolescents (13, 14).

The coordinated interregulation of lipid and carbohydrate metabolism includes control mechanisms such as the various peroxisome proliferator-activated receptors (PPAR) (15). Details regarding the nature of these pathways are still being investigated, although interventions targeting these mechanisms have therapeutic potential, as already seen with thiazolidinediones, agents that increase insulin sensitivity and decrease muscle and hepatic triglyceride load (16). DIDP and DINP have both been shown to act as peroxisome-proliferating agents, mainly through activation of PPAR-alpha, triggering a cascade of events leading to peroxisomal proliferation (17, 18). In contrast with consistent reports from experimental studies on rodents, the effect of PPAR-alpha agonists on insulin sensitivity in humans is less clear, possibly due to species-specific responsiveness of PPAR-alpha (19). Furthermore, available evidence from in vitro studies suggests that one of the possible mechanisms through which phthalates exposure may affect insulin sensitivity is oxidative stress, either by activation of peroxisome proliferator–activated receptors (20) or by changes in mitochondrial membranes potential and permeability (21). In turn, oxidative stress could contribute to the development of insulin resistance (22–24).

Published studies have identified strong correlations of HMW and DEHP metabolites with insulin resistance in adult males in the 1999–2002 National Health and Nutrition Examination Survey (NHANES) (25), association of phthalate metabolites with prevalent diabetes among women in the 2001–2008 NHANES (26), and association of bisphenol A and phthalates metabolites with risk of new onset type 2 diabetes in middle-age women (27). A further study from our group (28) identified an association between urinary DEHP concentrations and insulin resistance in adolescents in the 2003–2008 NHANES. Given that DINP and DIDP are increasingly used to replace DEHP, and given the context of increasing diabetes in youth globally (29, 30), concern about these newer environmental exposures as a possible contributor is warranted.

We performed a cross-sectional analysis of the 2009–2012 NHANES to examine associations of urinary phthalate metabolites with insulin resistance in adolescents. Relationship with LMW, HMW, DEHP, DINP, and DIDP were examined separately, as well as for individual metabolites.

Materials and Methods

NHANES is a continuous, multicomponent, nationally representative survey of the noninstitutionalized U.S. population administered by the National Centers for Health Statistics (NCHS) of the Centers for Disease Control and Prevention (CDC). Data from the questionnaire, laboratory, diet, and physical examination components were used in the present analysis, for which data are available in biennial groupings. Written consent, and child assent as appropriate, was obtained after approval by the National Center for Health Statistics Research Ethics Review Board. The New York University School of Medicine Institutional Review Board exempted this study from review because it is based on previously collected and deidentified data.

Measures of insulin resistance

We calculated homeostatic model assessment of insulin resistance (HOMA-IR) by multiplying fasting (>9-h fast) glucose (mmol/L) by fasting insulin (μU/mL) and dividing by 22.5 (31). It has strong correlation with the gold standard hyperinsulinemic-euglycemic patch-clamp test (31, 32). HOMA-IR was log (base e) transformed to account for skewed distribution, with results presented after retransformation to the original scale. To assess insulin resistance as a categorical outcome, we used the cut point of 4.39 units of HOMA-IR. This cut point was chosen based on a previous study that examined the distribution and associations of HOMA-IR with sex, race/ethnicity, age, and weight status, as measured by BMI, among U.S. adolescent in the NHANES. In this study by Lee et al (33) a HOMA-IR threshold of 2 SD above the mean for normal-weight adolescents with normal fasting glucose (>4.39) represented a conservative estimate derived from the lowest proportion of normal-weight children classified as having insulin resistance.

Our primary study outcome was insulin resistance, although we examined continuous HOMA-IR as an outcome in secondary analyses.

Measurement of urinary phthalates

Phthalates were measured in one spot urine sample, and analyzed using HPLC and tandem mass spectroscopy. More extensive methodological description is provided elsewhere (34). For phthalate concentrations below the level of detection [46.9% for mono-iso-nonylphthalate (MNP), 38.8% for mono-n-methylphthalate (MMP), 21.4% for mono-(2-ethylhexyl) phthalate (MEHP), 3.1% for mono-(3-carboxypropyl) phthalate (MCPP), <1% for all other metabolites], we substituted the limit of detection divided by the square root of 2, following common practice (14). To adjust for dilution, we included urinary creatinine as a covariate in all analyses, as suggested by CDC (35).

We grouped urinary biomarkers according to their use in product categories. Low molecular weight (LMW) phthalates (diethylpthalate, di-n-butylphthalate, and di-n-isobutylphthalate) are frequently added to personal care products to preserve scent (3), whereas HMW phthalates (DEHP, di-n-octylphthalate, and butylbenzylphthalate) are used as plasticizers for diverse applications (Table 1) (4). We calculated molar sums for LMW and HMW metabolites and, within the HMW category, molar sums for DEHP, DIDP, and DINP metabolites (Table 1), dividing by the creatinine (Cr) concentration of the sample to account for dilution. We expressed the LMW concentration as the sum of micromolar concentrations of MMP, monoethylphthalate (MEP), mono-butylphthalate (MBP), and mono-isobutylphthalate (MiBP). The HMW concentration was calculated as the sum of molarities of mono-(2-ethyl-5-carboxypentyl) phthalate (MECPP), MCPP, mono-(2-ethyl-5-hydroxyhexyl) phthalate (MEHHP), mono-(2-ethyl-5-oxohexyl) phthalate (MEOHP), MEHP, and monobenzylphthalate (MBzP). We calculated DEHP metabolite concentration by adding molarities of MEHP, MECPP, MEHHP, and MEOHP. The molarity of mono-carboxy-isononyl phthalate (MCNP) was used to represent DIDP concentration, as it is the main DIDP metabolite (36). DINP metabolite concentration was calculated by adding molarities of mono-carboxy-isooctyl phthalate (MCOP), a metabolite of DINP, MCPP, and mono-iso-nonylphthalate (MNP). There are alternative ways to calculate DINP metabolite concentration (using only MCOP and MNP); these would exclude MCPP because it is also a metabolite of di-n-butylphthalate, a LMW phthalate found in personal care products (37). For our main analyses, we calculated DINP concentration including MCPP but, in sensitivity analyses, we excluded MCPP to assess whether this influenced the results. Our primary exposure variables were log-transformed (to account for skewed distribution) micromolar concentrations of LMW, HMW, DEHP, DIDP, and DINP metabolites, although secondary analyses also examined tertiles, and analyzed individual metabolites to determine which metabolites are driving associations.

Potential confounders

Information on height and weight was based on measures taken by trained health technicians, who used data recorders and standardized procedures. We derived BMI z scores from 2000 Centers for Disease Control and Prevention (CDC) norms, incorporating height, weight, and sex; overweight and obese were categorized as BMI z score at least 1.036 and at least 1.64 (38).

Other measures came from surveys and laboratory assessments. Trained interviewers fluent in Spanish and English elicited total 24-hour calorie intake in person, using standard measuring guides to assist reporting of volumes and dimensions of food items (available on the CDC NHANES Web site). Calorie intake was examined as a continuous variable. Because exposure to tobacco smoke is a risk factor for metabolic syndrome in adolescence, we included serum cotinine in multivariable models. We categorized into low (<0.015 ng/mL), medium (<2 and ≥0.015 ng/mL), and high (≥2 ng/mL) categories (39).

Self-reported data on leisure-time physical activity (PA) obtained during household interviews was also included in our analysis. According to the Physical Activity Guidelines for Americans (PAGA), children and adolescents should engage in 60 minutes or more of PA each day (40). Reported minutes of moderate PA and minutes of vigorous PA were combined using the weighted approach recommended by PAGA, ie, 1 minute of vigorous PA is equivalent to 2 minutes of moderate PA, and examined as a continuous variable in our model.

Race/ethnicity was categorized into Mexican American, Other Hispanic, Non-Hispanic White, Non-Hispanic Black and Other, based on self report by 17–19-year-olds and caregiver report in 12–16-year-olds. Poverty-income ratio (PIR; family income divided by the federal poverty line for family size) was categorized into quartiles. To maximize sample size in multivariable analysis, “missing” categories were created for all potential confounders, except BMI category (n = 6). Except for PA (69.7% missing) and PIR (7.9%), less than 5% of values were missing for any confounding variable.

Statistical analysis

We conducted univariate and multivariate analyses, using Stata 12.0 (StataCorp), and following NCHS guidelines (41). Fasting samples and urinary phthalate measurements are collected from partially overlapping subsamples in NHANES. Although there are subsample weights for each, NCHS advises against use of either subsample weight, because sample sizes may get quite small when combining subsamples and weighted analyses may result in unstable and unreliable statistical estimates (R. Paulose, CDC, National Center for Health Statistics, personal communication, 2014). We therefore followed the practice of Stahlhut et al (25), performing unweighted analyses as our base analytic approach, and performed a series of sensitivity analyses.

LMW/HMW/DEHP/DINP/DIDP urinary metabolite concentrations were log transformed to account for skewed distribution, and divided into tertiles within the population with available phthalate and HOMA-IR data. We performed univariate regressions of logs of the molar concentrations of metabolite groups against HOMA-IR, insulin resistance, and each of the demographic, dietary, anthropometric, and the other covariates. We used multivariable linear regression analysis to model HOMA-IR, and logistic regression to model insulin resistance in separate models. In nested multivariable models we sequentially added: 1) urinary creatinine, 2) age and BMI category, 3) demographic and exposure characteristics (race/ethnicity, PIR, sex, serum cotinine), and 4) caloric intake and PA.

Sensitivity analysis

We assessed the robustness of our analysis by reprising 1) multivariable analyses using fasting weights, and 2) multivariable analyses using environmental sample weights. Given the high number of missing values for PA data, we also used multiple imputation techniques (43) to generate randomly a replacement value for each missing data point. We performed 20 imputations for the imputed variable following common practice (44). Specifically, we first imputed missing values of PA using univariate imputation with linear regression for continuous variables. We then reprised the full multivariable model using the “mi estimate” command in Stata version 12.0 to assess whether this affected relationships of log-transformed urinary phthalates with insulin resistance. The “mi” command checks the integrity of the estimation model (including consistency of estimation samples and omitted variables) and provides notification if a problem exists with the model.

We also tested the robustness of our results by including doctor-diagnosed diabetes and prediabetes covariates as well as kidney disease (estimated glomerular filtration rate <60 mL/min) and reprising multivariable results excluding adolescents with diabetes, prediabetes, and kidney disease. Because liver function may alter phthalate metabolism (45), we examined associations of metabolites with alanine aminotransferase (ALT) to rule out confounding by differences in liver function, and added categorized ALT in multivariable models of insulin resistance. In addition, we reprised multivariable results using alternative calculations of DINP that excluded MCPP.

Results

Of the 2612 children ages 12–19 years who participated, urinary phthalate metabolites were measured and available in 808 nonpregnant participants. Of these, fasting insulin/glucose were measured and available in 356 participants, who constituted our analytic sample. Median urinary LMW, HMW, DEHP, DINP, and DIDP metabolites were: 0.48 (interquartile range [IQR], 0.24–1.03; SE, 0.10), 0.35 (IQR, 0.16–0.73; SE, 0.05), 0.15 (IQR, 0.07–0.32; SE, 0.03), 0.07 (IQR, 0.03–1.18; SE, 0.02), and 0.008 (IQR, 0.004–0.02; SE, 0.002) micromolar. Median HOMA-IR was 2.80 (IQR, 1.88–4.23; SE, 0.09); 28.9% were insulin resistant (HOMA-IR ≥ 4.39). Five participants were classified as having kidney disease (estimated glomerular filtration rate <60 mL/min), three reported doctor-diagnosed diabetes, and one reported doctor-diagnosed prediabetes.

Descriptive analyses identified significantly higher levels of LMW metabolites among girls (Table 2), but not the other categories/metabolites. Whites had significantly lower LMW metabolite concentrations, whereas other Hispanics had higher DIDP metabolite concentrations. The category classified as “Other Race” (including multiracial) had significantly higher DINP metabolite concentrations. Adolescents from higher-income households had lower LMW metabolite concentrations. Adolescents with urinary cotinine concentrations between 0.015 and 1.9 ng/mL (58.1%) and those with excessive caloric intake (18.5%) had higher concentrations for DEHP. Finally, adolescents classified as insulin resistant (defined as HOMA-IR ≥ 4.39) had significantly higher concentrations of HMW, DEHP, and DINP metabolites.

Table 2.

Comparison of Urinary Phthalate Metabolites in Study Population With Fasting Insulin and Glucose Data in Pooled 2009–2012 NHANES (n = 356)

Characteristic	Total, N (%)	LMW		HMW		DEHP		DIDP		DINP
		Mean Urinary Metabolite, μM	Pa	Mean Urinary Metabolite, μM	Pa	Mean Urinary Metabolite, μM	Pa	Mean Urinary Metabolite, μM	Pa	Mean Urinary Metabolite, μM	Pa
Sex
Male	191 (53.7)	0.270	Reference	0.321	Reference	0.140	Reference	0.007	Reference	0.071	Reference
Female	165 (46.4)	0.415	<.001	0.341	.52	0.147	.62	0.008	.76	0.077	.53
Race/ethnicity
Hispanic–Mexican American	87 (24.4)	0.522	Reference	0.330	Reference	0.154	Reference	0.007	Reference	0.064	Reference
Hispanic–Other Hispanic	38 (10.7)	0.740	.054	0.414	.21	0.156	.96	0.010	.049	0.101	.052
Non-Hispanic White	87 (24.4)	0.358	.008	0.323	.89	0.148	.79	0.008	.46	0.076	.35
Non-Hispanic Black	95 (26.7)	0.578	.46	0.330	1.00	0.140	.53	0.007	.79	0.075	.36
Other	49 (13.8)	0.513	.92	0.454	.052	0.168	.63	0.008	.32	0.121	.003
PIR
First quartile (<0.83)	75 (21.1)	0.588	Reference	0.373	Reference	0.167	Reference	0.008	Reference	0.081	Reference
Second quartile (0.83–1.59)	72 (20.2)	0.517	.71	0.393	.73	0.180	.66	0.008	.99	0.079	.88
Third quartile (1.6–3.09)	98 (27.5)	0.504	.29	0.352	.69	0.143	.30	0.008	.98	0.078	.82
Fourth quartile (≥3.1)	83 (23.3)	0.411	.019	0.289	.081	0.126	.077	0.008	.88	0.072	.53
Missing	28 (7.9)	0.606	.88	0.399	.74	0.150	.63	0.008	.82	0.073	.19
Serum cotinine
<0.015 ng/mL	89 (25.0)	0.438	Reference	0.301	Reference	0.124	Reference	0.007	Reference	0.080	Reference
0.015–1.9 ng/mL	207 (58.1)	0.532	.11	0.375	.063	0.166	.018	0.008	.734	0.077	.81
≥2.0 ng/mL	60 (16.8)	0.530	.24	0.354	.31	0.145	.35	0.008	.56	0.088	.66
Missing	0 (0.0)
PA
Insufficient (<60 min/d)	1 (0.3)	0.778	Reference	0.162	Reference	0.087	Reference	0.002	Reference	0.035	Reference
Sufficient (≥60 min/d)	107 (30.1)	0.458	.58	0.363	.38	0.155	.56	0.008	.21	0.081	.51
Missing	248 (69.7)	0.528	.69	0.348	.41	0.145	.59	0.008	.21	0.080	.52
Caloric intake compared with needs in active child of age/sex
Appropriate	276 (77.5)	0.521	Reference	0.337	Reference	0.144	Reference	0.008	Reference	0.079	Reference
Excessive	66 (18.5)	0.438	.18	0.417	.094	0.192	.036	0.007	.92	0.081	.92
Missing	14 (3.9)	0.567	.75	0.356	.83	0.121	.52	0.008	.65	0.090	.70
Overweight status
Not overweight (<85th percentile)	215 (60.4)	0.491	Reference	0.339	Reference	0.642	Reference	0.007	Reference	0.079	Reference
Overweight (≥85th percentile)	135 (37.9)	0.520	.59	0.373	.35	0.155	.64	0.008	.17	0.083	.67
Missing	6 (1.7)	0.801	.22	0.327	.92	0.160	.84	0.007	.94	0.056	.51
Insulin status
Not resistant	253 (71.1)	0.485	Reference	0.320	Reference	0.137	Reference	0.007	Reference	0.073	Reference
Insulin resistant	103 (28.9)	0.563	.18	0.441	.003	0.192	.003	0.008	.23	0.101	.022
Missing	0 (0.0)

Total number of participants from some variables do not total to 356 due to missing data.

^aDerived using univariate regression of log molar concentration of urinary metabolites. Mean urinary phthalate metabolites represents retransformed mean from log base.

Associations of phthalate metabolites with insulin resistance outcomes

Table 3 examines association of LMW, HMW, DEHP, DIDP, and DINP metabolites with insulin resistance (logistic regression modeling) and HOMA-IR as a continuous measure (linear regression modeling). In linear regression, for each log unit increase in HMW metabolites, a 0.09 increase in HOMA-IR was identified (P = .027), whereas for each log unit increase in DINP metabolites, a 0.08 increase in HOMA-IR (P = .008) was identified (Model A, controlling for urinary creatinine only). These associations were confirmed in multivariable models that controlled incrementally for BMI category and age; sociodemographic factors; caloric intake, and PA in the entire sample, and identified consistent relationships of HMW, DEHP, and DINP urinary metabolites with insulin resistance outcomes. Indeed, in Model D, for each log unit increase in HMW and DINP metabolite, 0.07 (P = .025) and 0.08 (P = .001) increases in HOMA-IR were identified. In logistic regression, in the final multivariable model (Model D), significantly higher odds for insulin resistance in the study population were associated with log-transformed urinary HMW (odds ratio [OR], 1.71; confidence interval [CI], 1.24–2.37; P = .001), DINP (OR, 1.39; CI, 1.09–1.76; P = .007), and DEHP metabolites (OR, 1.72; CI, 1.27–2.34; P = .0001).

Table 3.

Linear and Logistic Regression Analysis of Insulin Resistance Outcomes Associated With Urinary Phthalate Metabolites

	Model A (n = 356)	Model B (n = 350)	Model C (n = 350)	Model D (n = 350)
Increment, HOMA-IR
Log-transformed LMW Metabolite	+0.03 (−0.04, +0.10)	+0.02 (−0.04, +0.08)	+0.009 (−0.06, +0.07)	+0.01 (−0.05, +0.08)
Log-transformed HMW Metabolite	+0.09 (+0.01, +0.16)a	+0.07 (+0.007, +0.13)a	+0.07 (+0.009, +0.14)a	+0.07 (+0.009, +0.14)a
Log-transformed DEHP Metabolite	+0.07 (−0.003, +0.14)	+0.06 (+0.004, +0.12)a	+0.06 (+0.002, +0.12)a	+0.06 (−0.001, +0.12)
Log-transformed DIDP Metabolite	+0.06 (−0.01, +0.14)	+0.03 (−0.03, +0.01)	+0.04 (+0.03, +0.10)	+0.03 (−0.03, +0.01)
Log-transformed DINP Metabolite	+0.08 (+0.02, +0.13)b	+0.07 (+0.02, +0.12)b	+0.08 (+0.03, +1.12)b	+0.08 (+0.03, +0.13)b
OR, Insulin Resistance
Log-transformed LMW Metabolite	1.18 (0.93, 1.50)	1.30 (0.96, 1.76)	1.28 (0.92, 1.78)	1.28 (0.92, 1.79)
Log-transformed HMW Metabolite	1.47 (1.14, 1.89)b	1.67 (1.23, 2.27)b	1.72 (1.25, 2.36)b	1.71 (1.24, 2.37)b
Log-transformed DEHP Metabolite	1.42 (1.12, 1.80)b	1.68 (1.26, 2.24)c	1.73 (1.28, 2.34)c	1.72 (1.27, 2.34)c
Log-transformed DIDP Metabolite	1.17 (0.91, 1.50)	1.14 (0.84, 1.56)b	1.16 (0.85, 1.60)b	1.16 (0.84, 1.60)
Log-transformed DINP Metabolite	1.24 (1.03, 1.50)a	1.35 (1.08, 1.70)b	1.38 (1.09, 1.75)b	1.39 (1.09, 1.76)b

Ten participants with missing poverty-income data are insulin resistant and these are included in the analysis.

HOMA-IR categorized using cutpoint of 4.39.

Model A controls for urinary creatinine. Model B adds age and BMI category to Model A. Model C adds sex, PIR, serum cotinine, and race/ethnicity to Model B. Model D adds caloric intake and and PA to Model C.

Results using unweighted modeling are presented. See Supplemental Appendix for testing of alternative weighting.

^aP < .05.

^bP < .01.

^cP < .001.

These associations and their significance can be more easily identified when the study population is categorized into tertiles (Table 4): adolescents with the highest DINP concentrations (0.23 μm; 95% CI, 0.08–0.38; P = .003) had higher HOMA-IR than the others. In particular, when considering insulin resistance as a categorical outcome, an OR of 2.33 in the third tertile (participants with the highest concentration of urinary DINP metabolites) translates into an adjusted prevalence for insulin resistance of 34.4% (95% CI, 27.3–41.6%; P = .033), compared with a prevalence of 23.4% in the first tertile. Similarly, compared with the first tertile of DEHP (20.5% adjusted prevalence), the third tertile had 37.7% prevalence (95% CI, 29.8–45.6%; P = .003); compared with the first tertile of HMW (18.8% adjusted prevalence), the second tertile had 33.8% prevalence (95% CI, 25.6–41.9; P = .006), and the third tertile had 36.7% prevalence (95% CI, 29.3–44.1%; P = .001).

Table 4.

Linear and Logistic Regression Analysis of Insulin Resistance Outcomes Associated With Urinary Phthalate Metabolites (Tertiled)

	LMW Metabolites	HMW Metabolites	DEHP Metabolites	DIDP Metabolites	DINP Metabolites
Model D, Increment, HOMA-IR (n = 350) Predicted HOMA-IR
First Tertile	Reference	Reference	Reference	Reference	Reference
	1.08 (0.96–1.20)	0.99 (0.88–1.10)	1.02 (0.91–1.13)	1.05 (0.95–1.15)	0.99 (0.89–1.10)
Second Tertile	+0.08 (−0.08, 0.24)	0.18 (0.01, 0.35)a	0.07 (−0.09, 0.23)	0.06 (−0.08, 0.21)	0.07 (−0.09, 0.22)
	1.16 (1.06–1.26)	1.17 (1.06–1.28)	1.09 (0.98–1.19)	1.11 (1.00–1.22)	1.06 (0.95–1.16)
Third Tertile	−0.03 (−0.21, 0.15)	0.14 (−0.03, 0.30)	0.16 (−0.005, 0.33)	0.08 (−0.08, 0.25)	0.23 (0.08, 0.38)b
	1.05 (0.94–1.15)	1.13 (1.03–1.23)	1.18 (1.07–1.30)	1.13 (1.01–1.25)	1.22 (1.12–1.32)
Model D, OR, Insulin Resistance (n = 350) Prevalence
First Tertile	Reference	Reference	Reference	Reference	Reference
	24.8.% (17.5–32.2)	18.8% (13.2–24.5)	20.5% (14.5–26.5)	26.0% (19.6–32.3)	23.4% (17.2–29.6)
Second Tertile	1.91 (0.86–4.28)	3.51 (1.43–8.63)b	2.32 (0.98–5.51)	1.58 (0.75–3.34)	1.63 (0.72–3.68)
	33.3% (26.2–40.6)	33.8% (25.6–41.9)	30.5% (23.0–38.0)	32.1% (24.7–39.5)	29.4% (21.9–36.9)
Third Tertile	1.35 (0.53–3.44)	4.30 (1.81–10.22)b	3.85 (1.60–9.24)b	1.32 (0.56–3.13)	2.33 (1.07–5.07)a
	28.6% (21.3–36.0)	36.7% (29.3–44.1)	37.7% (29.8–45.6)	29.6% (21.3–38.0)	34.4% (27.3–41.6)

Increases are per log unit in urinary LMW/HMW/DEHP/DIDP/DINP metabolite concentration. See Methods for calculation.

Tertiles for LMW: <0.275 μM (n = 106); 0.275–0.696 μM (n = 118); ≥0.696 μM (n = 132). Tertiles for HMW: <0.238 μM (n = 127); 0.238–0.535 μM (n = 104); ≥0.535 μM (n = 125). Tertiles for DEHP: <0.105 μM (n = 130); 0.105–0.241 μM (n = 113); ≥0.241 μM (n = 113). Tertiles for DIDP: <0.006 μM (n = 146); 0.006–0.013 μM (n = 107); ≥0.013 μM (n = 103). Tertiles for DINP: <0.045 μM (n = 124); 0.045–0.120 μM (n = 110); ≥0.120 μM (n = 122).

All models control for continuous urinary creatinine, age, and caloric intake as well as sex, PIR, serum cotinine, BMI, race/ethnicity categories, and PA.

Results using unweighted modeling are presented.

^aP < .05.

^bP < .01.

The examination of individual phthalate metabolites identified associations of HOMA-IR and insulin resistance with two metabolites of DINP, monocarboxyisooctyl phthalate (MCOP) and MNP, in linear and logistic regression modeling (Table 5). In linear regression modeling, for each log unit increase in MCOP and MNP, a 0.07 and 0.09 increments in HOMA-IR (P = .003 and P = .0001, respectively) were identified; for these two metabolites of DINP the association was also present in logistic regression modeling as significantly higher odds for insulin resistance. No significant association was identified for MCNP, a metabolite of DIDP.

Table 5.

Associations of Individual Urinary Phthalate Metabolites With Insulin Resistance Outcomes in Linear and Logistic Regression Analyses

Increment/OR	Increment, HOMA-IR (n = 350)	OR, Insulin Resistance (n = 350)
Low-molecular weight phthalates
Log-transformed MEP	+0.02 (−0.05, +0.05)	1.18 (0.90, 1.54)
Log-transformed MBP	+0.05 (−0.02, +0.12)	1.80 (1.21, 2.69)b
Log-transformed MiBP	+0.03 (−0.05, +0.11)	1.74 (1.16, 2.59)b
Log-transformed MMP	−0.02 (−0.007, +0.03)	1.05 (0.82, 1.35)
High-molecular weight metabolites (non-DEHP)
Log-transformed MBzP	+0.03 (−0.03, +0.09)	1.34 (0.98, 1.83)
Log-transformed MCPP	+0.08 (+0.03, +0.13)b	1.37 (1.07, 1.76)a
Log-transformed MCOP	+0.07 (0.02, +0.11)b	1.34 (1.07, 1.68)a
Log-transformed MNP	+0.09 (0.04, +0.13)c	1.45 (1.15, 1.82)b
Log-transformed MCNP	+0.03 (−0.03, +0.10)	1.16 (0.84, 1.60)
High-molecular weight metabolites of DEHP
Log-transformed MEHP	+0.04 (−0.02, +0.09)	1.52 (1.15, 2.00)b
Log-transformed MEHHP	+0.05 (−0.0005, +0.11)	1.63 (1.24, 2.15)b
Log-transformed MEOHP	+0.05 (−0.004, +1.11)	1.68 (1.25, 2.26)b
Log-transformed MECPP	+0.05 (−0.01, +0.11)	1.60 (1.19, 2.14)b

Increases are per log unit in urinary LMW/HMW/DEHP/DINP/DIDP metabolite concentration. See Methods for calculation.

All models control for continuous urinary creatinine, age and caloric intake as well as sex, PIR, serum cotinine, BMI, race/ethnicity categories, and PA.

Results using unweighted modeling are presented.

^aP < .05.

^bP < .01.

^cP < .001.

Log-transformed urinary concentrations of the HMW, non-DEHP metabolite MCPP (P = .002) were significantly associated with increases in HOMA-IR (Table 5). In logistic regression, for DEHP metabolites, log-transformed MEHP (OR, 1.52; P = .003), MEHHP (OR, 1.63; P = .001), MEOHP (OR, 1.68; P = .001), MECPP (OR, 1.60; P = .002), were significantly associated with insulin resistance. Associations with insulin resistance were also present for two LMW phthalates metabolites, MBP (OR, 1.80; P = .004), and MiBP (OR, 1.74; P = .007) (Table 5).

In our sensitivity analyses, we first excluded adolescents previously diagnosed with prediabetes, diabetes, and kidney disease, and this did not change the results (data not shown). We then included prediabetes, diabetes, and kidney disease (categorized as present/absent) in our model, and inclusion of these covariates did not change the associations of HOMA-IR and insulin resistance with urinary HMW/DINP metabolites (data not shown). Multiple imputation of PA data and using alternative calculations of DINP urinary concentrations also did not change the results (data not shown).

A single association with categorized, but not continuous, ALT was identified only for log-transformed total DEHP metabolites (OR, 1.74; 95% CI, 1.07–2.81; P = .025). Given this finding, we reprised our multivariable model with the inclusion of ALT as a covariate, as well as exclusion of adolescents with elevated ALT (≥30 U/L). The significance of the association of insulin resistance with DEHP metabolites was maintained with less than 10% of attenuation, whereas significance of the association with DINP metabolites was unchanged (Supplemental Table 2).

Discussion

In this study we identify, for the first time, an association between urinary DINP metabolites and insulin resistance in adolescents; we also confirm the findings of a previous study from our group (28) showing the association of DEHP metabolites with insulin resistance. The association between DINP and insulin resistance is confirmed whether urinary phathalate metabolites are modeled as continuous or categorical variables, and remained significant after the inclusion of a rich set of information regarding demographics, exposures, and lifestyle variables, thus providing more convincing evidence for nonspuriousness.

Relationships between phthalate (including DEHP, DIDP, and DINP) intake and urinary metabolites are complex and, to our knowledge, no pharmacokinetic studies in children/adolescents have been performed.

After exposure, LMW phthalates are hydrolyzed in the corresponding monoesters and eliminated without further metabolism. However, for those phthalates with long alkyl chains (HMW), including DEHP, DINP, and DIDP, the monoesters are further metabolized into oxidative products (through ω- and ω-1-oxidation of the monoester side chain), which are more water soluble and easily eliminated through urinary excretion (46, 47).

For DINP, the corresponding hydrolytic monoester is monoisononyl phthalate (MNP), which is then further metabolized to oxidized metabolites, including monocarboxyisooctyl phthalate (MCOP) (48). With respect to DIDP, this is first hydrolyzed to monoisodecylphthalate then metabolized to oxidized metabolites, such as monocarboxyisononyl phthalate (MCNP) (49). Current evidence suggests that, especially with respect to DINP, measuring only MNP underestimates human exposure, whereas oxidized metabolites such as MCOP are more sensitive indicators of exposure, with higher frequency of detection and higher urinary concentrations (36). For this reason, in the present study, we investigated not only the relationship of MNP with insulin resistance outcomes but also the relationship of the corresponding oxidative metabolite MCOP, and both were found to be significantly associated with insulin resistance outcomes. DIDP metabolites were not found to be associated with insulin resistance in this study, and one of the possible explanations is the lower urinary concentrations of DIDP identified in this sample, compared with DINP. This might be related to different sources of exposure: while DINP is used in plastic products for food packaging, making diet the most likely source of exposure in adolescents, DIDP is predominantly used in furnishings, cookware, medications, and other consumer products (50), to which adolescents could be less exposed.

Although LMW phthalates have been associated with obesity among children and adolescents (14, 51), such evidence is not available for HMW/DINP metabolites, suggesting that exposures may act independently of increased body mass to produce insulin resistance but further research is needed to interrogate mechanisms.

PPARs play key roles in lipid and carbohydrate metabolism (52) and a plausible mechanism for the association of DEHP metabolites with insulin resistance is activation of PPAR-gamma receptor transcription (53). With respect to DINP, available evidence suggests that it stimulates both PPAR-alpha and PPAR-gamma activation, but a lower activation concentration is required for the former (54). PPAR-alpha plays an important role in lipid catabolism and fatty acids oxidation (15) (55) and activation of PPAR-alpha seems to increase mitochondrial fatty acid oxidation (52). In turn, increased fatty acid oxidation has been reported to be a source of reactive oxygen species in the heart (56), skeletal muscle (57), and kidney cortical tubules (58), suggesting a potential mechanism through which DINP, by increasing production of reactive oxygen species, could contribute to the development of insulin resistance (59).

Even when coupled with potential mechanisms (activation of PPAR receptors, oxidative stress) causation cannot be inferred from a cross-sectional study. Many potential confounders were unmeasured, including pubertal status data, collected in 2011–2012 NHANES but not currently available. Phthalates may affect pancreatic function earlier in life than the adolescent period we examined in the present study, and longitudinal studies of prenatal and infant exposure are needed. Another alternative explanation would be that insulin-resistant children have unhealthy eating behaviors including more packaged food consumption, and thus have higher urinary levels.

Phthalate exposure is measured at one time point in this analysis, and monoesters of phthalates are more typically known to have half-lives of 12–48 hours (60), yet phthalates are known to deposit in fat, thereby lengthening half-life beyond that identified in the few adult pharmacokinetic studies (61). Urinary phthalates are more likely to represent current as opposed to chronic exposure and temporal variability is expected to be different among the various compounds. We currently have no data on DINP and temporal variability over time but, even if current urinary phthalates are weak indices of exposure, our estimates of association should be biased toward the null for dichotomous outcomes (62).

Knowledge gaps also persist in understanding sources of exposure for DINP, although available data suggest that food is the main source (approximately 60–70%) in adults, teens, and children (63–65). These findings were incorporated in the recently published Chronic Hazard Advisory Panel (CHAP) report, released by the U.S. Consumer Product Safety Commission in 2014, which recommended permanently banning DINP (66). Although dietary sources are likely to be the chief source of exposure, given the uses of DINP in other products, we cannot rule out nondietary sources as contributors to the associations identified here. Evidence in support of the notion that phthalate exposure contributes to insulin resistance and other metabolic disruption is increasing, and there are ample opportunities for policy prevention in the current regulatory void, especially in light of the potential associated health costs, as evidenced in a recently published analysis on the burden and disease costs of exposure to endocrine disrupting chemicals in the European Union (42).

Conclusion

We have identified a cross-sectional association of DINP and DEHP metabolite exposure with increased insulin resistance in a sample of U.S. adolescents. Additional studies are warranted both to confirm the association and elaborate the different potential mechanisms involved.