Endocrine Disruptors and Asthma-Associated Chemicals in Consumer Products

Background: Laboratory and human studies raise concerns about endocrine disruption and asthma resulting from exposure to chemicals in consumer products. Limited labeling or testing information is available to evaluate products as exposure sources. Objectives: We analytically quantified endocrine disruptors and asthma-related chemicals in a range of cosmetics, personal care products, cleaners, sunscreens, and vinyl products. We also evaluated whether product labels provide information that can be used to select products without these chemicals. Methods: We selected 213 commercial products representing 50 product types. We tested 42 composited samples of high-market-share products, and we tested 43 alternative products identified using criteria expected to minimize target compounds. Analytes included parabens, phthalates, bisphenol A (BPA), triclosan, ethanolamines, alkylphenols, fragrances, glycol ethers, cyclosiloxanes, and ultraviolet (UV) filters. Results: We detected 55 compounds, indicating a wide range of exposures from common products. Vinyl products contained > 10% bis(2-ethylhexyl) phthalate (DEHP) and could be an important source of DEHP in homes. In other products, the highest concentrations and numbers of detects were in the fragranced products (e.g., perfume, air fresheners, and dryer sheets) and in sunscreens. Some products that did not contain the well-known endocrine-disrupting phthalates contained other less-studied phthalates (dicyclohexyl phthalate, diisononyl phthalate, and di-n-propyl phthalate; also endocrine-disrupting compounds), suggesting a substitution. Many detected chemicals were not listed on product labels. Conclusions: Common products contain complex mixtures of EDCs and asthma-related compounds. Toxicological studies of these mixtures are needed to understand their biological activity. Regarding epidemiology, our findings raise concern about potential confounding from co-occurring chemicals and misclassification due to variability in product composition. Consumers should be able to avoid some target chemicals—synthetic fragrances, BPA, and regulated active ingredients—using purchasing criteria. More complete product labeling would enable consumers to avoid the rest of the target chemicals.


CONTENTS:
S-2 Table S1. Chemical class summaries, including uses, potential health effects, and compounds analyzed S-5 Figure S1. Concentrations of target compounds in sunscreen samples S-6 Figure S2. Correlation matrix for conventional and alternative products S-7 Figure S3. Effects of compositing products on detection patterns: An example using 5 alternative sunscreens.
S-8 Figure S4. Number of products per composite versus number of detected chemicals for conventional samples.

S-9
Analytical Quality Assurance/Quality Control (QA/QC) Methods and Results S-11 Table S2. Summary statistics for solvent method blanks (µg) in each analytical round.

S-18 References cited in Supplemental Material
S-2 Table S1. Chemical class summaries, including uses, potential health effects, and compounds analyzed.

Analytical Quality Assurance/Quality Control (QA/QC) Methods and Results
To composite samples: for thin liquids, a 1 ml aliquot of each sample was combined and mixed, then 0.25 mL was removed and spiked with the surrogate recovery standards (SRSs), diluted with 50 mL of 3:1 dichloromethane (DCM):methanol, ultra-sonified for 1 min and placed on a shaker table for 15 min; for thick liquids (e.g., toothpaste) and semi-solid materials (e.g., lipstick), approximately 1 g of each product was added in sequence to a tared vial and then mixed with a spatula; 0.25 g was removed, spiked with the SRSs, mixed with 0.5 g muffled Extrelute to form a free-flowing mixture, ultra-sonified and shaken in 50 mL 3:1 DCM:methanol, and filtered as needed through a syringe filter; for solids, equal amounts of each product were finely divided, mixed and 0.25 g was removed, spiked with the SRSs, and extracted as described above. A 6 mL aliquot of each extract was passed through a weak anion exchange SPE cartridge (DSC-NH2, 100 mg; Supelco), 1 mL was removed, spiked with the neutral internal standard (IS; bromobiphenyl), and analyzed using GC/MS in the full scan mode for neutral compounds. A separate aliquot of the extract was solvent exchanged into the same volume of methyl-t-butyl ether, 1 mL was removed, spiked with the IS dibromophenol-d3 and 50 uL pyridine, derivatized with BSTFA with 1% TMCS, and then analyzed using GC/MS in the full scan mode for phenolic compounds. The same phenolic extract was reanalyzed in the multiple ionization detection mode for the multi-component nonylphenol mono-and diethoxylates. The internal standard method of quantification was based on 6 point calibration curves that spanned the range of 0.15-25 µg/mL.
QA/QC measures were conducted to ensure accuracy and reliability of measurements. To evaluate potential contamination, we analyzed eight solvent method blanks in round one and five solvent method blanks in round two. To estimate precision we analyzed eight samples in round one and five samples in round 2 as duplicates. Surrogate recoveries were used to characterize accuracy and extraction efficiency.
For each compound, the method reporting limit (MRL) was defined as the maximum of the analytical detection limit and the 90 th percentile of the blank concentrations within each analytical round. The nominal analytical detection limit was 1 ug/g, and so the MRL was above 1 ug/g only in cases with detectable concentrations in the blank samples. Concentrations above the MRL were considered quantified and presented in the paper.
In order to correct potential bias in the reported values, blank correction was performed for chemicals detected in at least 75% of the blanks and were specific to each analytical round. In analytical round one, none of the compounds required blank correction. In analytical round two, one compound (D5; decamethylcyclopentasiloxane) was blank corrected by subtracting the median blank concentration from the reported values, resulting in a median 87% change and maximum 200% change to reported concentration.

S-10
Precision is presented as the relative percent difference between duplicate pair concentrations (Table  S3). Summary statistics were calculated only using pairs of detected values. In general, for the 28 analytes with pairs of detected values, precision estimates were <50%, except in 2 instances (ethyl paraben and 4-nonylphenol). There were 7 analytes with "mismatched" pairs, meaning that one of the duplicate pairs was detected and one was not.
Five surrogate compounds (octyl alcohol-d17, 4-chlorophenyl phenyl ether-d5, di-n-butyl phthalate-d4, phenanthrene-d10 and bisphenol A-d16) were spiked into samples prior to extraction at 1 µg/ml level (or 200 µg/g; levels of octyl alcohol-d17 increased to 2 µg/ml in the second analytical round for increased sensitivity) to evaluate accuracy and extraction efficiency. Median percent recoveries were within the 50-150% acceptance range for all surrogates over both analytical rounds ( Figure S5). There were sporadic recoveries outside of this range in the 71 Round 1 samples for bisphenol A-d16 (below 50% in 5 samples and above 150% in 6 samples), and octyl alcohol-d17 (below 50% in 23 samples and above 150% in 1 sample). Variable recoveries may be attributable to interferences from complex matrices. Reported concentrations were not adjusted or corrected.
Six products were spiked prior to extraction with the full suite of analytes at concentrations of 100 ng/ml for the alkylphenol ethoxylates and 2.5 µg/ml for all other compounds, and analyzed to assess recovery. Three types of conventional and alternative products were selected to represent different matrices: glass cleaner (liquid), sunscreen (cream), and scrubbing powder (semi-solid). Thirteen recovery estimates (of 406 possible) could not be estimated due to matrix interference. Median recoveries across all products were within 50-150%, except for benzophenone-2 (not detected in any samples), monoethanolamine, and diethanolamine, which had low recoveries (Table S4). Therefore, the results presented for these compounds may be underestimates of the true concentrations. There were no significant differences in recoveries across product type (p>0.05).
S-11 Table S2. Summary statistics for solvent method blanks (µg/g equivalent a ) in each analytical round.    pinene  fragrances  8  0  0  0  0  1  5  0  0  0  0  1  terpineol  fragrances  8  0  0  0  0  1  5  0  0  0  0  1  AHTN  fragrances  8  0  0  0  0  1  5  0  0  0  0  1  bucinal  fragrances  8   NA 5 0 0 0 0 1 NA indicates "not applicable" since compound was added during second analytical round MRL represents the method reporting limit, which is the maximum of the 90th percentile of the blanks or the analytical detection limit (1 ug/g). MRLs have been blank corrected when necessary. a Calculated assuming 50 ml extract volume and 0.25 g sample, which was used for all product samples b Subjected to blank correction by subtracting median blank concentration from reported values S-13 Table S3. Summary statistics for precision, calculated as relative percent difference (%).