Adolescent exposures to cosmetic chemicals of concern
Teen Girls' Body Burden of Hormone-Altering Cosmetics Chemicals: Study methodology
20 dedicated teen volunteers started a morning between April and September 2007 not with classes, sports, or a summer job, but with a 50 mL blood draw at a local clinic. This diverse group of young women shared a common curiosity – the desire to learn about chemical contaminants in their bodies, a pollution they may have absorbed through use of personal care products.
The potential for increased vulnerability during adolescence to hormone disruption and other harmful effects of contaminants, combined with the tendency of teen girls to expose themselves to more, and more varied, cosmetic ingredients through use of body care products, suggests that special attention be directed to assessing chemical exposures for this group. Motivated to fill current data gaps, EWG analyzed blood and urine samples from 20 teens for ingredients commonly found in cosmetics and personal care products – 25 chemicals in total, from 4 chemical classes. To our knowledge this work includes the first reported tests of teens for 17 of these chemicals. Information below describes the components of this new study, detailing the sample collection procedures, sample preparation and analysis methods, the quality assurance and quality control provisions included in the study design, and the analysis of associations between contaminant levels and specific personal care products used. Study materials and protocols were reviewed by Alan Greene, MD FAAP, a Clinical Professor of Pediatrics at Stanford University School of Medicine, an Attending Pediatrician at Packard Children's Hospital, and a Senior Fellow at the University California San Francisco Center for the Health Professions. Study design and implementation were approved by an institutional review board that reviewed ethical and data quality considerations of the investigation.
In collaboration with the Teens for Safe Cosmetics coalition and its umbrella organization, Search for the Cause, EWG recruited 20 teens willing to provide blood and urine samples for analysis. The teens ranged in age from 14 to 19, and came from all over the U.S., representing Northern, Central, and Southern California, Florida, Massachusetts, New Jersey, Oregon, Pennsylvania, Texas, Washington, and Washington DC. The diverse pool of participants included Caucasian (14), African American (3), Asian American (2), and multi-ethnic (1) girls.
Sample collection and storage
Each participant received a sample kit, allowing her to collect her first morning void of urine at home, and then proceed to a local phlebotomy clinic for a 50 mL blood draw. Clinic staff processed the blood to extract the serum, then shipped urine and serum samples to Axys Analytical Services, British Columbia. Axys stored samples at -20 degrees Celsius, and analyzed urine samples for phthalates, triclosan, and parabens. Axys shipped serum samples to TNO Laboratories, The Netherlands, where they were analyzed for musks. Three serum samples were damaged in transit; musk measurements could not be made for these samples.
Analysis of phthalates
Phthalate metabolites were extracted from 1 mL urine samples. Urine samples were first buffered with ammonium acetate, and then spiked with 13C-labeled phthalate monoesters, 13C4-4-methylumbelliferone, and 4-methylumbelliferyl glucuronide. Urine samples were also spiked with beta-glucuronidase enzyme (for deconjugation of glucuronidated forms of the target analytes). The treated samples were incubated to hydrolyze the glucuronides (the completeness of hydrolysis was monitored by the recovery of 4-methylumbelliferone).
Samples were extracted and purified using solid phase extraction (SPE) procedures. Extracts were spiked with labeled recovery standards for phthalate metabolites and analyzed by liquid chromatography tandem mass spectrometry (LC/MS/MS) using a Micromass Quattro Ultima MS/MS coupled with a Waters 2695 HPLC system. The method determined the total of the free and the glucuronidated phthalate metabolites. Analyte concentrations were determined using isotope dilution quantification. Values are reported in units of micrograms per gram creatinine, a urine protein, to account for variation in the dilution of the urine samples due to different levels of fluid intake by the participants.
Analysis of triclosan
Triclosan metabolites were extracted from 4 mL urine samples. Urine samples were first buffered with ammonium acetate buffer and then spiked with a labeled triclosan standard solution, 13C4-4-methylumbelliferone, and 4-methylumbelliferyl glucuronide, and beta-glucuronidase enzyme (for deconjugation of possible glucuronidated forms of the target analyte). The treated samples were incubated to hydrolyze the glucuronides (the completeness of hydrolysis was monitored by the recovery of 4-methylumbelliferone).
Samples were extracted and purified using SPE procedures employing Waters Oasis Max cartridges. Extracts were spiked with labeled recovery standards and analyzed by LC/MS/MS using Micromass Quattro Ultima MS/MS coupled with a Waters 2695 HPLC system. The method determined the total of the free and the glucuronidated triclosan metabolites. Analyte concentrations were determined using isotope dilution quantification. Values are reported in units of micrograms per gram creatinine, a urine protein, to account for variation in the dilution of the urine samples due to different levels of fluid intake by the participants.
Analysis of nitro- and polycyclic musks
Serum samples were analyzed by TNO Laboratories, The Netherlands, in compliance with NEN-EN-ISO/IEC 17025 and RvA accreditation no. 54, “The development and application of methods for the determination of organic contaminants in environmental matrices, wastes and material.” TNO Environment and Geosciences Laboratories is listed in the RvA register under no. L 026. RvA is the Dutch Council for Accreditation and is a member of the European cooperation for Accreditation (EA) and the International Laboratory Accreditation Cooperation (ILAC). In addition TNO Environment and Geosciences Laboratories operates in compliance with the Quality System standard ISO 9001.
All glassware used in the analysis for artificial musks was cleaned, rinsed with demineralized water and baked in an oven for 16 hours at 280°C prior to use. All solvents were distilled prior to use to achieve low blank results
For each analysis, a serum sample is placed in a 60 mL clean glass vial. Methanol, 0.1 M HCl and a set of internal standards were added to the sample. The sample was extracted twice with a hexane-diethyl ether mixture and centrifuged after each extraction to separate the organic phase. The combined extracts were washed with a 1% KCl solution and dried with anhydrous sodium sulphate.
The extract was concentrated to a small volume and purified over a glass chromatography column packed with partially deactivated florisil. The fraction containing the artificial musk compounds was collected and concentrated. A syringe standard was added and the final extract was analyzed with gas chromatography coupled with mass spectrometry (GC/MS) in the selected ion monitoring mode (SIM).
The identification of analytes was based on correct retention times and qualifier ion ratios compared to external standards analyzed together with the sample extracts. The quantification was based on a response factor determined from the external standards. The recovery of the added internal standards was used to determine the performance of the method but not to correct the results. Results are expressed in parts per billion parts serum wet weight.
Analysis of parabens
Paraben metabolites were extracted from 2 mL urine samples. Urine samples were spiked with deute-rium labeled quantification standards, 4-methylumbelliferyl glucuronide, 4-methylumbelliferyl sulfate solution, and then buffered with ammonium acetate. Samples were subjected to enzymatic hydrolysis using beta-glucuroni-dase/sulfatase enzyme for deconjugation of conjugated glucuronide and sulfonate forms of the target analytes (the completeness of hydrolysis was monitored by the recovery of 4-methylumbelliferone).
Samples were extracted and purified using SPE procedures employing Waters HLB cartridges. Extracts were spiked with labeled recovery standards and analyzed by LC/MS/MS using a Micromass Quattro Ultima MS/MS coupled with a Waters 2695 HPLC system. The method determined the total of the free and the glucuronidated triclosan metabolites. Analyte concentrations were determined using isotope dilution quantification.
Values are reported in units of micrograms per gram creatinine, a urine protein, to account for variation in the dilution of the urine samples due to different levels of fluid intake by the participants.
Procedures for quality assurance and quality control (QA/QC)
To assess the possibility that air or dust in the home setting could contribute to the levels of contaminants in participants’ urine, some participants were asked to collect “field blank” samples, exposing jars of purified water to the sampling (bathroom) environment simultaneous to collection of the urine sample. EWG staff performed a similar field blank sample collection procedure during collection and processing of serum samples for these participants to assess the hospital environment as a source of additional contamination. A single set of field blanks was analyzed, and no evidence of contamination from home or hospital air was detected.
All urine analyses were conducted in accordance with AXYS' accredited QA/QC program. Regular participation in international inter-laboratory calibration programs is a component of this program. Each analysis batch included a procedural blank to demonstrate cleanliness, a spiked laboratory control sample to monitor precision and recovery, a batch duplicate to monitor precision and a batch un-spiked matrix (USM) using reference synthetic urine to determine the levels of analytes prior to spiking. The sample results were reviewed and evaluated in relation to the QA/QC samples processed at the same time. The sample surrogate standard recoveries and detection limits, procedural blank data and the laboratory control sample data were evaluated against method criteria to ensure acceptable data quality.
The serum analyses were conducted in accordance with TNO’s ISO-9001 accredited QA/QC program. Upon receipt, the samples were checked for integrity and spiked with isotope labeled internal standards to facilitate compound identification. The recovery of these standards (musk-xylene-d15 and tonalide-d3) is used to determine the performance of the analytical procedure, but not used to correct the results, as these standards are not compound specific.
As part of the QA/QC, control samples were injected and analyzed for each series of real samples. Control samples consisted of a similar matrix or duplicate sample with standard addition of the analytes, or alternatively, the analysis of a duplicate sample. Initial quality control procedures included: assessment of the performance of the instrumental analysis system; assessment of method blank samples and external standards; visual inspection of the chromatograms on peak separation and interferences; identification criteria, e.g. retention time and isotope ratios; quantification criteria, e.g. number of scans/peak, recovery of the internal/surrogate standard, recovery of the standard addition in control samples and minimal signal/noise ratios; and assessment of raw data and calculations by a second technician. The final results of any analysis are validated against the performance criteria of each method before they are reported.
Analysis of personal care products
On the day of the blood draw, participants were asked to complete a questionnaire detailing the personal care products they used, including brand names and other identifying information. Participants were asked how frequently they used the products, and for those products used within the last 48 hours, the number of hours elapsed between their previous 2 uses and the blood draw. 19 of 20 participants provided this information.
EWG researchers linked questionnaire responses to our Skin Deep cosmetics database, an online database of more than 32,000 personal care products, which includes lists of ingredients and associated health concerns. Using this information, we analyzed participants’ body levels of cosmetic chemicals against the ingredients in all products they used, as well as in products used daily or during the 24 hours prior to the blood draw.
50% of all products listed in questionnaire responses were matched exactly to products found in Skin Deep; 91% of products used daily or within 24 hours of sampling were matched exactly. For products that could not be matched to any found in Skin Deep, probabilistic modeling was used to assign products present in the database that were of the same product type (e.g. lotion or sunscreen). 30 iterations of this probabilistic model provided suitable exposure statistics concerning the number of products used containing cosmetic chemicals of concern.
Extensive analysis revealed no statistically significant correlations between participants’ contaminant levels and recent (daily or within 24 hours) or overall product usage for each chemical under study. In addition, no correlations were noted between body levels and exposures for chemical families, or for body levels of musks, phthalates, or diethyl phthalate specifically and exposures to “fragrance” in cosmetics. This lack of correlation is not surprising, given the small number of participants, the lack of information about both the concentrations of each ingredient within body care products and the expected amount of uptake with use, as well as the multiple non-cosmetic sources of exposures to these chemicals that commonly occur.