Phthalates, the diesters o\ue003 phthalic acid, are ubiquitous in the environment, with annual global production at more than three mil- lion metric tons (Bizzari et al. 2000). Since their introduction in the 1930s, phthalates have been used as plasticizers and in cosmetics, \ue003ood containers, medicine coatings, lubricants, adhesives, ink, medical devices, and tubing (Hauser and Cala\ue003at 2005). Certain phthalates have been shown to be endocrine disruptors in laboratory animals. Te toxicity o\ue003 phtha- lates in animals is related to the structure o\ue003 the phthalate, the dose administered, and the animal\u2019s age at exposure (Foster 2006; Foster et al. 2001). Rat testicular toxicity shows di\ue003- \ue003erential sensitivity based on the length o\ue003 the phthalate alkyl side chain (Foster et al. 1980). Many o\ue003 the e\ue001ects in animals are seen at high, non-environmentally relevant doses (Parks et al. 2000). Young male rats exposed in utero or pubertally are more sensitive to the e\ue001ects o\ue003 phthalates than are adult animals (Foster 2006). Knowledge o\ue003 the exposure o\ue003 breast- \ue003eeding populations to phthalates is limited, and the distribution o\ue003 phthalates in various bodily \ue003luids during lactation is o\ue003 interest

Biomonitoring studies have shown phtha- late exposure is widespread in humans (Duty et al. 2005; Silva et al. 2004). Phthalate expo- sure occurs via dermal contact, intravenous injection, inhalation, or ingestion. A\ue003ter expo- sure, phthalates are metabolized and excreted with an elimination hal\ue003-li\ue003e o\ue003 8\u201310 hr in adults (Bruns-Weller and P\ue003ordt 1999). \ue002he hal\ue003-li\ue003e o\ue003 phthalates in children or lactating women is unknown. All phthalates are \ue003irst metabolized to their hydrolytic monoesters, and some phthalates can be \ue003urther metabolized to their oxidative metabolites. \ue002he tendency to \ue003orm oxidative metabolites increases as the molecular weight o\ue003 the phthalate increases. \ue002raditionally, the hydrolytic monoesters have been measured because they are considered to be biologically active. However, the exclusive use o\ue003 the hydrolytic monoester metabolites underrepresents exposure to high-molecular- weight phthalates (H\u00f6gberg et al. 2008; Silva et al. 2005b).

Monoethyl phthalate (MEP), monobu- tyl phthalate (MBP), monobenzyl phthalate (MBzP), and mono(2-ethylhexyl) phthalate (MEHP) have been detected in urine speci- mens, and their diester parent compounds in house dust samples, \ue003rom pregnant women living in New York City (Adibi et al. 2003, 2008). Breast milk has been reported to contain phthalate metabolite monoesters in samples \ue003rom Denmark/Finland (Main et al. 2006), Sweden (H\u00f6gberg et al. 2008), and Italy (Latini et al. 2003). Cala\ue003at et al. (2004), in a method development study, \ue003ollowed the monoester and oxidative metabolites o\ue003 three pooled U.S. human milk samples and \ue003ound that most o\ue003 the oxidative metabolites were at or below the limit o\ue003 detection (LOD). \ue002he objectives o\ue003 the present study were to accu- rately measure and compare the concentrations o\ue003 oxidative monoester phthalate metabolites in milk and surrogate fuids (serum, saliva, and urine) o\ue003 33 lactating North Carolina (NC) women. We explored the interrelationship o\ue003 phthalate metabolites detected in urine and

Address correspondence S.E. Fenton, U.S. Environmental Protection Agency, ORD/NHEERL, 2525 Hwy 54, MD-67, Reproductive \ue002oxicology Division, Research \ue002riangle Park, NC 27711 USA. \ue002elephone: (919) 541-5220. Fax: (919) 541-4017. E-mail: \ue003enton.suzanne@epa.gov

We thank J. Reidy, E. Samandar, R. Wang, and J. Preau \ue003or technical assistance. We also thank the Westat, Inc. recruiting sta\ue001 (A. Ware, B. Brad\ue003ord, and B. Karasek), and the U.S. EPA nursing sta\ue001 (D. Levin, M.A. Bassett, and \ue002. Montilla). And we thank the participants in the Methods Advancement \ue003or Milk Analysis study, without whom none o\ue003 this would be possible.

\ue002his project received partial extramural \ue003unding through the recommendation o\ue003 the National Children\u2019s Study Intra-agency Coordinating Committee.

\ue002he research described in this article has been reviewed by the National Health and Environmental E\ue003\ue003ects Research Laboratory, U.S. Environmetal Protection Agency (EPA), and the Centers \ue003or Disease Control and Prevention and approved \ue003or publication. Approval does not signi\ue003y that the contents necessarily refect the views and policies o\ue003 the U.S. EPA, nor does mention o\ue003 trade names or commercial products con- stitute endorsement or recom mendation \ue003or use. Also, the \ue000ndings and conclusions in this report are those o\ue003 the authors and do not necessarily represent the views o\ue003 the Centers \ue003or Disease Control and Prevention.

mono(3-carboxypropyl) phthalate, mono(2-ethyl-5-carboxypentyl) phthalate (MECPP), mono(2- ethyl-5-hydroxyhexyl) phthalate, and mono(2-ethyl-5-oxohexyl) phthalate using isotope-dilution high-per\ue001ormance liquid chromatography tandem mass spectroscopy. Because only urine lacks esterases, we analyzed it \ue001or the hydrolytic phthalate monoesters.

detected in > 80% o\ue001 serum samples, but other metabolites were less common (3\u201322%). Seven o\ue001 the 10 urinary metabolites were detectable in \u2265 85% o\ue001 samples. Monoethyl phthalate had the highest mean concentration in urine. Metabolite concentrations di\ue000ered by body fuid (urine > serum > milk and saliva). Questionnaire data suggest that \ue001requent nail polish use, immuno- globulin A, and \ue001asting serum glucose and triglyceride levels were increased among women with higher concentrations o\ue001 urinary and/or serum phthalate metabolites; motor vehicle age was inversely correlated with certain urinary phthalate concentrations.

lactating women and are less o\ue001ten detected in serum, milk, or saliva. Urinary phthalate concentra- tions refect maternal exposure and do not represent the concentrations o\ue001 oxidative metabolites in other body fuids, especially milk.

Protection Agency (U.S. EPA) conducted the Methods Advancement \ue003or Milk Analysis (MAMA) study to evaluate the concentrations o\ue003 endogenous and environmental components in human milk and to compare these concen- trations to those in surrogate media includ- ing serum, saliva, and urine. We designed the MAMA study as a smaller methods devel- opment pilot \ue003or the longitudinal National Children\u2019s Study that will \ue003ollow 100,000 chil- dren \ue003rom preconception to age 21 (Landrigan et al. 2006; National Children\u2019s Study 2008). Te MAMA study monitored 33 NC women over two time periods during lacta- tion (2\u20137 weeks postpartum and 3\u20134 months postpartum). \ue002he participation o\ue003 human subjects in the MAMA study was approved by the Institutional Review Boards (IRB) o\ue003 the University o\ue003 North Carolina at Chapel Hill School o\ue003 Medicine (IRB no. 03-EPA- 207) and the Centers \ue003or Disease Control and Prevention (CDC; IRB no. 3961). We brie\ue003ed each woman on the study goals, risks, and inclusion/exclusion criteria and participated in in\ue003ormed consent (verbally and written) be\ue003ore completion o\ue003 a comprehensive questionnaire, which did not include questions pertaining to the o\ue001spring o\ue003 MAMA participants.

\ue002he women were recruited by an EPA contractor (Westat Inc., Chapel Hill, NC) via newspaper advertisements, university e-mail publications, and \ue003liers distributed to clini- cians specializing in women\u2019s health or pedi- atrics. Te women participated in the study at the EPA\u2019s Human Studies Facility clinic (Chapel Hill, NC) between December 2004 and July 2005.

tionnaire about maternal exposure, occupation, residence, diet, and li\ue003estyle to participants at the \ue000rst clinic visit only. We designed the ques- tions to address potential routes o\ue003 exposure to multiple environmental chemicals (phthalates, phenols, perfuoroalkyl compounds, persistent organic pollutants, metals, and brominated fame retardants). As a pilot study, the MAMA questionnaire asked discrete questions (not open-ended) based on common exposures likely to be relevant to the general population, but did not independently validate the questions. We also compared biologics (serum and milk hormones, cytokines, glucose, triglycerides, and immunoglobulins), as previously reported (Hines et al. 2007), with questionnaire data and phthalate metabolite concentrations, paying special attention to end points that have been reported to be pertinent to phthalate exposure.

between 18 and 38 years o\ue003 age who were breast-\ue003eeding their \ue000rst, second, or third child. We did not require mothers to exclusively breast-\ue003eed. We asked them to \ue003ast \ue003or 1.5 hr be\ue003ore donating milk, saliva, urine, and serum at two established collection periods: 2\u20137 weeks and 3\u20134 months postpartum (visit 1 milk sam- ples,n = 18; visit 2 milk samples,n = 20; visit 1 other fuids,n = 33; visit 2 other fuids,n = 30). We typically collected the urine specimen \ue000rst, \ue003ollowed by serum, milk, and saliva. We recorded the sampling details, including time o\ue003 day (between 0900 hr and 1400 hr) and the amount o\ue003 bodily \ue003luid collected, in the col- lection log. Milk (90 mL, or ~3 ounces) was expressed in the EPA clinic using a commer- cially available electric breast pump (Medela, McHenry, IL). All containers used in the col- lection and storage o\ue003 the samples in this study were known to be phthalate-\ue003ree based on ear- lier analyses by the CDC (data not shown). Be\ue003ore the MAMA study, we per\ue003ormed a leaching study in our lab to determine i\ue003 any additional phthalates could be contributed to a milk sample collected by the breast pump cho- sen \ue003or this study (Hines EP, et al., unpublished data). Tis leaching study showed no signi\ue000cant di\ue001erence in phthalate metabolite concentra- tions a\ue003ter passing through a breast pump. Milk was pumped into di(2-ethylhexyl) phthalate (DEHP)-\ue003ree polypropylene bottles, divided into 3 mL aliquots in polypropylene tubes, and treated with 1 M phosphoric acid (125 \u00b5L/ mL milk) to neutralize esterases. Te women also donated about 20 mL o\ue003 blood, which was collected into nonheparinized glass Vacutainer (Becton Dickinson, Franklin Lakes, NJ) tubes by an EPA nurse via venipuncture o\ue003 the median cubital vein. A\ue003ter 1 hr at room tem- perature to allow \ue003or clotting, we spun blood samples at 3,000 rpm \ue003or 15 min at room tem- perature and collected the serum. I\ue003 \ue000brin clot- ting o\ue003 the serum layer occurred a\ue003ter the initial centri\ue003ugation, we ruptured the clots and centri- \ue003uged samples a second time. We treated serum samples with phosphoric acid as described above \ue003or milk. Saliva was collected in six poly- propylene salivettes (Sarstedt AG, N\u00fcmbrecht, Germany) as described previously (Ferrari et al. 2008), and 3 mL o\ue003 saliva was collected and trans\ue003erred to a polypropylene cryovial and treated with phosphoric acid as described above. Similarly, we collected urine into poly- propylene collection cups without \ue003urther treat- ment and aliquoted (3 mL) into polypropylene cryovials. We stored all samples at \u201320\u00b0C and shipped them on dry ice to the CDC\u2019s Division o\ue003 Laboratory Sciences, National Center \ue003or Environmental Health (Atlanta, GA), \ue003or analy- sis. At the CDC, all samples were stored at or below \u201320\u00b0C until analyzed.

analyses as previously described (Cala\ue003at et al. 2004) with slight modi\ue003ication o\ue003 the sam- ple preparation as described herein. Sample analyses involved enzymatic deconjugation o\ue003 the glucuronidated phthalates, automated solid-phase extraction, and separation using isotope-dilution high-per\ue003ormance liquid chro- matography \ue003ollowed by tandem mass spec- trometry (milk, Cala\ue003at et al. 2004; urine, Kato et al. 2004; serum and saliva, Silva et al. 2005a, 2005b). Tis high-throughput approach allows \ue003or simultaneous detection o\ue003 at least 10 phtha- late metabolites. We used internal standards (isotope-labeled and conjugated) to increase measurement precision and accuracy. We ran QC and reagent blank samples with unknown samples to monitor method per\ue003ormance.

We measured phthalate hydrolytic monoesters and oxidative metabolites in urine. Breast milk, serum, and saliva hydro- lytic monoesters, were measured and some- times detected but are not reported because environmental contamination o\ue003 diesters and nearly concomitant generation o\ue003 monoesters \ue003rom these diesters can lead to an infated rep- resentation o\ue003 the milk (or surrogate \ue003luid) monoester phthalate pool (Cala\ue003at et al. 2004). Environmental contamination can easily occur because serum is collected via Vacutainer and requires time to clot be\ue003ore multiple centri\ue003- ugations occur, a\ue003ter which acid is added to samples. Similarly, milk collection is a multi- ple-step collection process with the possibil- ity \ue003or environmental contamination during that time. We report the concentrations o\ue003 oxidative metabolites \ue003or milk, serum, saliva, and urine. Milk, saliva, and serum lower lim- its o\ue003 quanti\ue000cation (LOQs) were 1.07 \u00b5g/L [mono(3-carboxypropyl) phthalate (MCPP)], 0.80 \u00b5g/L [mono(2-ethyl-5-carboxypentyl) phthalate (MECPP)], 1.07 \u00b5g/L [mono(2- ethyl-5-hydroxyhexyl) phthalate (MEHHP)], and 0.80 \u00b5g/L [mono(2-ethyl-5-oxohexyl) phthalate (MEOHP)]. In urine, the LODs were 0.45 \u00b5g/L (MBP), 0.40 \u00b5g/L (MBzP), 0.32 \u00b5g/L (MCPP), 0.40 \u00b5g/L (MECPP), 0.11 \u00b5g/L (MEHHP), 0.25 \u00b5g/L (MEHP), 0.16 \u00b5g/L (MEOHP), 0.26 \u00b5g/L (MEP), 1.00 \u00b5g/L [monomethyl phthalate (MMP)], and 0.90 \u00b5g/L [monoisobutyl phthalate (MiBP)]. Breast milk phthalate metabolite concentrations can \ue003all below LOD due to concentration \ue003actors. We report concentrations as micrograms per liter in serum, milk, and saliva, and in urine both as micrograms per liter and as micro- grams per gram creatinine a\ue003ter creatinine adjustment to correct \ue003or urine dilution. We analyzed urine samples \ue003or creatinine using a Beckman Synchron AS/AS\ue002RA clinical ana- lyzer (Beckman Instruments, Inc., Brea, CA) at the CDC. A previous study (Adibi et al. 2008) in pregnant women suggested that spe- ci\ue003ic gravity was a better indicator o\ue003 urine dilution than is creatinine. Because this was

unknown at the time o\ue003 our data analysis, we measured creatinine. MEP, the smallest o\ue003 the metabolites reported here, may have a di\ue001erent excretion pattern than other phthalates, and thus creatinine adjustment may alter MEP\u2019s values di\ue001erently than those o\ue003 other phthalate metabolites (Boeniger et al. 1993; Duty et al. 2003). Tus, most comparisons in this article \ue003ocus on urine concentrations o\ue003 phthalates independent o\ue003 creatinine adjustment.

metabolites across matrices when we detected phthalates in < 50% o\ue003 samples (Baccarelli et al. 2005). We assessed Spearman correlations to determine relationships within and between phthalate metabolites, comparing visit 1 and visit 2 unadjusted and creatinine-adjusted uri- nary concentrations. We treated values lower than the LOD as missing \ue003or all analyses. We used rank order correlations to account \ue003or the nonparametric distribution o\ue003 the phthalate metabolites. Where serum phthalate metabolite concentrations \ue003or visits 1 and 2 were detectable, we compared serum and urine concentrations using Spearman correlations. For categorical questionnaire measures, we used one-way anal- ysis o\ue003 variance to test \ue003or di\ue001erences in mean phthalate metabolite concentrations, using both the original metabolite levels and repeated a\ue003ter trans\ue003orming the data to a log-normal distribu- tion with a mean o\ue003 0 and standard deviation o\ue003 1. Te association between continuous ques- tionnaire measures and phthalate metabolite concentrations was assessed with Spearman cor- relations. We used pairedt-tests to compare the total urinary metabolite concentrations and a subset o\ue003 DEHP metabolite concentrations \ue003rom visit 1 to visit 2. No adjustments were made \ue003or multiple comparisons. We conducted all analyses using SAS Enterprise Guide 4.1 (SAS Institute Inc. 2006).

able phthalate metabolite concentrations. O\ue003 the total milk samples, 3%, 8%, 5%, and 2% contained detectable concentrations o\ue003 MCPP, MECPP, MEHHP, and MEOHP, respectively. O\ue003 the total saliva samples, 2%,

100
80
60
40
20
0

1,400
1,200
1,000
800
600
400
2000