Determination of phthalate monoesters in human milk, consumer milk, and infant formula by tandem mass spectrometry (LC–MS–MS)
- First Online:
- Cite this article as:
- Mortensen, G.K., Main, K.M., Andersson, A. et al. Anal Bioanal Chem (2005) 382: 1084. doi:10.1007/s00216-005-3218-0
Daily exposure of humans to phthalates may be a health risk because animal experiments have shown these compounds can affect the differentiation and function of the reproductive system. Because milk is the main source of nutrition for infants, knowledge of phthalate levels is important for exposure and risk assessment. Here we describe the development and validation of a quantitative analytical procedure for determination of phthalate metabolites in human milk. The phthalate monoesters investigated were: monomethyl phthalate (mMP), monoethyl phthalate (mEP), mono-n-butyl phthalate (mBP), monobenzyl phthalate (mBzP), mono-(2-ethylhexyl) phthalate (mEHP), and monoisononyl phthalate (mNP). The method is based on liquid extraction with a mixture of ethyl acetate and cyclohexane (95:5) followed by two-step solid-phase extraction (SPE). Detection and quantification of the phthalate monoesters were accomplished by high-pressure liquid chromatography using a Betasil phenyl column (100 mm×2.1 mm×3 μm) and triple tandem mass spectrometry (LC–MS–MS). Detection limits were in the range 0.01 to 0.5 μg L−1 and method variation was from 5 to 15%. Analysis of 36 milk samples showed that all these phthalates were present, albeit at different concentrations. Median values (μg L−1) obtained were 0.11 (mMP), 0.95 (mEP), 3.5 (mBP), 0.8 (mBzP), 9.5 (mEHP), and 101 (mNP). We also analysed seven samples of consumer milk and ten samples of infant formula. Only mBP and mEHP were detected in these samples, in the ranges 0.6–3.9 μg L−1 (mBP) and 5.6–9.9 μg L−1 (mEHP).
KeywordsPhthalatesExposureAnalysisHuman milkConsumer milkInfant formula
Phthalates (diesters of phthalic acid) are industrial chemicals used in a variety of flexible plastic products, as additives to poly(vinyl chloride) (PVC), and in consumer items ranging from personal care (nail polish, lotions, and perfumes), pharmaceutical coatings, paints, and insect repellent to toys and food packaging materials. Thus phthalates are ubiquitously present in our environment and human exposure can occur both via ingestion, inhalation, or dermal routes [1–4].
Several phthalates have been shown in animal models to be developmental and reproductive toxicants, because of their endocrine disrupting effects [5–11]. The metabolism of phthalates does not usually detoxify the compounds and experiments have shown that DBP and DEHP were metabolised to bioactive phthalate monoesters [5, 6, 12]. Hydrolysis of the phthalate diesters to the corresponding monoesters usually occurs in both kidney and liver, but phthalates with short chains, for example dimethyl phthalate (DMP) and dibutyl phthalate (DBP), the most water soluble compounds, are more readily metabolised than di-(2-ethylhexyl) phthalate (DEHP) [13–15]. Glucuronidation of the monoester or further degradation of the residual alkyl chain increases water solubility and facilitates excretion [13, 16, 17]. In human urine approximately 71% of mEP is excreted as its free monoester whereas the more lipophilic monoester mEHP was excreted primarily as the glucuronidated form; only approximately 15% was excreted as the free monoester . Urinary analysis has demonstrated widespread human exposure to phthalates. Results from the National Health and Nutrition Examination Survey (NHANES) 1999–2000 show that females in The United States have higher concentrations of mBP than males but similar mEHP levels . Of particular interest, children aged 6–11 years seem to have higher levels of mBP, mBzP, and mEHP than adults .
Because milk is a major source of nutrition for infants the presence of phthalates in human milk is of increasing interest. During the past 20–30 years persistent chemicals, for example organochlorine pesticides, PCB, dioxins, and organobromine contaminants have been measured in human milk [20–22]. The presence of these persistent chemicals reflects environmental contamination and, because of regulation, levels of chlorinated compounds have decreased whereas those of brominated compounds have increased . There are few reports on levels of phthalates in human milk, although in a few pooled samples phthalate monoesters were measured in the range 1 to 16 μg L−1 . Because human milk is a complex matrix consisting of both proteins and different amounts of fat , contaminants including phthalates can bind to these matrix constituents with different affinity, depending on their chain lengths [13, 26].
The main purpose of this study was to develop and validate a new quantitative analytical procedure for determination of phthalate monoesters in human milk using LC–MS–MS. Although general adult exposure to phthalates is to diesters, phthalate monoesters were chosen because maternal diester intake is metabolised to monoesters. The described method was used for analysis of 36 milk samples collected in a contemporary prospective Danish baby cohort. Samples of consumer milk and infant formulas were also analysed.
Study population and collection of milk samples
Milk from healthy mothers (36 samples) was randomly chosen from a prospective, longitudinal baby cohort study performed from 1997 to 2001 at the National University Hospital in Copenhagen, Denmark. Pyrex glass bottles (1515/06D, 250 mL; Bibby Sterilin, Staffordshire, UK) with Teflon coated caps were given to the mothers at birth with oral and written instructions to collect milk from months 1 to 3 postnatally. The baby was fed first, then milk aliquots were collected into a glass or porcelain cup avoiding, if possible, the use of breast pumps. Milk was frozen in the glass bottle as additive aliquots and delivered frozen to the hospital. Here it was stored at −20°C until analysis. This sampling procedure was chosen to ensure that breastfeeding had been well established and the baby was thriving.
The study was conducted according to the Helsinki II declaration and approved by the ethical committee (KF01-030/97) and the Danish Data Protection agency (1997-1200-074).
Consumer milk and infant formula
Seven samples of common consumer milk from organic and conventional farming with fat content ranging from 1.5 to 3.5% were collected. Ten samples of infant formula products were collected; two of the formula products were ready to use, the remaining samples were powders that were reconstituted according to the instructions given by the manufacturers in standard milk bottles used in Danish homes. Formula products were based on milk from both organic and conventional farming.
All phthalate monoesters (>98% purity): monomethyl phthalate (mMP), monoethyl phthalate (mEP), mono-n-butyl phthalate (mBP), monobenzyl phthalate (mBzP), mono-(2-ethylhexyl) phthalate (mEHP), and monoisononyl phthalate (mNP) and their 13C4-labelled internal standards were purchased from Cambridge Isotope Laboratories (Andover, MA, USA) as solutions (100 μg mL−1 in acetonitrile (ACN)). Working solutions were prepared at concentrations at 10 μg mL−1 in water–ACN (1:1). Unlabelled monoester spike solution (0.1–3.0 μg mL−1), labelled monoester internal spike solution (0.1–2.0 μg mL−1), and solutions for calibration curves in the range 0.0025 to 0.5 μg mL−1 were prepared in water–ACN (9:1).
All reagents were analytical or HPLC grade and Milli-Q water was cleaned in a Millipore system (Synthesis A10). Acetonitrile, methanol, and sodium sulfate, anhydrous, were supplied by J.T. Baker (Deventer, Holland), ethyl acetate was from BDH Laboratory Supplies (Poole, England). Cyclohexane, acetic acid 100%, phosphoric acid 85%, ammonia solution 25%, and sodium hydrogen phosphate dihydrate (NaH2PO4.2H2O) were obtained from Merck (Darmstadt, Germany). 4-Methylumbelliferone, 4-methylumbelliferyl β-D-glucuronide, and ammonia acetate 7.5 mol L−1 were supplied by Sigma–Aldrich (Steinheim, Germany). β-Glucuronidase (Escherichia coli K12) was purchased from Roche Diagnostics (Mannheim, Germany). All reagents and water were checked for contamination with phthalate monoesters before use.
Two buffers were prepared for sample preparation. Acidic buffer: NaH2PO4.2H2O (10.9 g) and phosphoric acid (5 mL) were dissolved in 500 mL Milli-Q water. Basic buffer: ammonia solution (0.5 mL) was added to 100 mL water–ACN (1:1).
Human serum and milk both contain enzymes which are involved in the metabolism of phthalate diesters to phthalate monoesters, and also phthalate diesters that may be present because of ex vivo contamination of the samples, can be degraded [24, 27]. Therefore, phosphoric acid is typically added to serum samples to stop the degradation of possible contaminating phthalate diesters in the sample . To test that contamination with phthalate diesters during handling in the laboratory did not contribute to the measured monoester levels milk samples were thawed and phosphoric acid was added to half of the sample at the same ratio as used for serum . Milk with and without acid was stored at room temperature and at −18°C. Samples were analysed at start and after 1, 4, and 24 h.
Analytical methods used for determination of environmental contaminants in human milk often include liquid extraction then clean-up procedures depending on the specific compounds and the final measurement technique [20–22, 28–30]. Removal of matrix compounds is important to obtain a final extract which can be injected into the analytical system without contamination problems and hence, at the same time, achieve high sensitivity (signal-to-noise ratio). Based on an existing method for extraction of pesticides from milk  we tested different mixtures of ethyl acetate, ethanol, and cyclohexane for extraction of phthalate monoesters. A 95:5 mixture of ethyl acetate and cyclohexane was selected and sodium sulfate was added as drying agent. Alkaline digestion of human milk has been used to reduce association of bisphenol A and 4-nonylphenol with milk proteins , but this procedure degrades the phthalate monoesters.
The final extraction mixture consisted of milk (3 mL), acidic buffer (3 mL), extraction mixture (15 mL) and sodium sulfate (6 g). With this extraction procedure it was possible to obtain a clear extract for further clean up. Milk samples were thawed and placed in a water bath at 37°C to furnish a homogeneous sample without a separate fat layer. Internal standards (75 μL, 0.1–2.0 μg mL−1) were added to the extraction mixture. The tubes were shaken 10 min and centrifuged at 4000 rpm for 5 min. Supernatant (10 mL) was evaporated to dryness and the residue was dissolved in 1.0 mL basic buffer for further clean-up and analysis. To include the glucuronide-bound fraction β-glucuronidase enzyme (15 μL in 750 μL ammonia acetate 1 mol L−1) was added to the milk (3 mL) and the samples were incubated at 37°C for 90 min for deglucuronidation. 4-Methylumbelliferyl β-D-glucuronide (1 mg L−1, 60 μL) was added to the sample before incubation as a control for enzymatic deconjugation.
Solid-phase extraction (SPE)
Different solid-phase extraction (SPE) methods have been reported for phthalates in urine and serum [27, 31, 32]; a method using SPE without a liquid extraction has also been reported . Adjusting the solution to basic or acidic conditions is often used to separate matrix constituents from the analytes. Thus, we tried to purify the liquid extract using one acidic clean up and tested both wash procedures and elution steps. Including a washing step with methanol caused loss of the most polar phthalate monoesters, whereas washing with water or acidic buffer was insufficient to clean the solution. We therefore decided to clean the liquid extract using two SPE cartridges (Oasis HLB, Waters, Milford, MA, USA) based on a urine phthalate extraction method . The first SPE cartridge was used to retain hydrophobic compounds while the more acidic phthalates were eluted. An SPE cartridge (60 mg, 3 mL) was equilibrated with ACN then with basic buffer. The residue from the liquid extraction dissolved in basic buffer was eluted and then acidified with acidic buffer (3 mL). A new SPE cartridge (200 mg, 6 mL) was equilibrated with ACN then with Milli-Q water and acidic buffer. The acidified eluate was added to the cartridge, which was washed with acidic buffer (3 mL) and Milli-Q water (3 mL). The phthalates were eluted with ACN (2 mL) then ethyl acetate (2 mL) and the extract was evaporated to dryness under a gentle stream of nitrogen at 42°C. The residue was resuspended in 500 μL water–ACN (9:1) and transferred to a glass autosampler vial insert for analysis.
Instrumentation and conditions
The HPLC gradient programme used to separate the actual phthalate monoesters
Per cent A (0.1% acetic acid in water)
Per cent B (0.1% acetic acid in acetonitrile)
MS instrument settings
A Finnigan TSQ Quantum Ultra triple quadrupole mass spectrometer in combination with Xcalibur software was used for detection and quantification (Thermo Electron Corporation, San Jose, CA, USA). The instrument was run in negative-ion mode using the electrospray source (ESI). Spray voltage was set to 5,000 V. Sheath and auxiliary gas pressures were optimised to 30 and 8 psig, respectively. Capillary temperature was set to 350°C. Ion source collision-induced dissociation (CID) was experimentally selected during development of the method by measuring signal-to-noise ratios. It was shown that an increase of the internal energy of the ions by applying source CID resulted both in good signals and in good signal-to-noise ratios and for the actual matrix; an optimum was obtained at 10 V.
Monitored ions, collision energies, tube lens settings, and retention times for the measured phthalate monoesters: monomethyl phthalate (mMP), monoethyl phthalate (mEP), mono-n-butyl phthalate (mBP), monobenzyl phthalate (mBzP), mono-(2-ethylhexyl) phthalate (mEHP), and monoisononyl phthalate (mNP)
Collision energy (V)
Tube lens offset (V)
Retention time (min)
Recovery experiments at two different levels ranging from 2 μg L−1 to 120 μg L−1 were included using different milk samples. Native standards (60 or 120 μL, 0.1–3.0 μg mL−1) were added to the sample before extraction. The mean coefficient of variation (CV, %) was calculated using the concentration differences between real duplicate determinations of different milk samples. Determination of the repeatability (r) and reproducibility (R), i.e. the variation within and between days , was determined on a pooled milk sample without any spike additions. The sample was analysed four times on four different days.
Blank samples were run routinely and always measured before and after the milk samples.
Validation data for the new method
Validation data for: monomethyl phthalate (mMP), monoethyl phthalate (mEP), mono-n-butyl phthalate (mBP), monobenzyl phthalate (mBzP), mono-(2-ethylhexyl) phthalate (mEHP), and monoisononyl phthalate (mNP)
Detection limit (μg L−1)
Repeatability (r) (%)
Reproducibility (R) (%)
Recovery low level (%)
Recovery high level (%)
Variation CV (%)
Recoveries at both spiking levels were between 93% and 100% (Table 3). The deviations were highest for mBP, mEHP and mNP because of higher background levels for these compounds, but still acceptable for determination at low levels in a complicated matrix.
The repeatability (r) was between 4.8% and 14% whereas the reproducibility (R) was between 4.4% and 18%. The coefficient of variation (CV, %) describing the difference between duplicate determinations of different samples was between 4.8% and 16%.
Monoester concentrations in human milk
Concentrations of free phthalate monoesters in human milk
Minimum (μg L−1)
Maximum (μg L−1)
Median (μg L−1)
Mean (μg L−1)
Standard deviation (μg L−1)
Half of the samples were analysed after enzymatic deglucuronidation, to measure the total (free and glucuronidated) phthalate monoester content. The results showed that all the investigated phthalate monoesters were present almost exclusively in the free form (69–97%).
The fat content of the samples was determined at The National Public Health Institute, Kuopio . The amounts were in the range 0.47% to 4.8% with a mean fat content of 2.9±1.1%. The median concentrations of phthalate monoesters calculated as μg g−1 fat were 0.004 (mMP), 0.03 (mEP), 0.12 (mBP), 0.03 (mBzP), 0.4 (mEHP), and 3.5 (mNP).
For 13 of the milk samples breast pumps had been used during the sampling. Levels of phthalate monoesters in samples collected without the use of breast pump were compared with levels in samples collected with use of a breast pump by means of the Mann–Whitney U-test. There was no significant difference between levels of mMP, mBzP, mEHP, or mNP in samples collected with or without a breast pump. Levels of mEP and mBP were, however, significantly higher in the samples that had been collected by use of breast pumps (mEP median level 1.48 compared with 0.56 μg L−1, P=0.007 and mBP median level of 5.06 compared with 1.95 μg L−1, P=0.03).
Monoester concentrations in consumer milk and infant formula
Concentration ranges for phthalate monoesters in consumer milk and infant formula, including recoveries and variations obtained for these products
Detection limit (μg L−1)
Infant formula (μg L−1)
Consumer milk (μg L−1)
Recovery low level (%)
Recovery high level (%)
Variation CV (%)
We here present a new method for quantitative determination of phthalate monoesters in human milk. The detection limits obtained were comparable with or lower than those reported for other environmental contaminants including phthalates in milk [24, 28, 29] or phthalates in urine and serum [27, 32] and therefore sufficient for quantification of phthalate monoesters in the milk. The recovery and repeatability obtained show that the method is robust and reliable for this complex matrix at the measured concentration levels.
Using this method we measured phthalate monoesters in 36 individual human milk samples. All six tested phthalate monoesters were present in all the samples, except for mMP, which was below the detection limit in approximately 10% of the samples (4 out of 36). Large inter-individual variation was observed with a skewed distribution and few samples with high concentrations. Although all the phthalates are used in a variety of consumer products the most water-soluble phthalate diesters (DMP and DEP) are used mostly in personal care products. Di-butyl phthalate (DBP) is also used in cosmetics whereas the more lipophilic butyl benzyl phthalate (BBzP), di-(2-ethylhexyl) phthalate (DEHP), and di-iso-nonyl phthalate (DNP) are used primarily for industrial purposes as plasticisers. Personal behaviour can, therefore, presumably affect exposure to, e.g., DEP and DBP more than exposure to DEHP and DNP; this may explain why the largest inter-individual variation in human milk content was observed for mBP and mEP (the metabolites of DBP and DEP), including the very high concentrations of mEP and mBP in a few of the samples. DBP is also used in pharmaceutical coatings and intake with medication has previously been implicated as an explanation of a very high concentration of mBP in a sample of urine . Currently, however, we have no evidence of a specific exposure source which can explain the large intra-individual variation in mBP observed in our samples.
Levels of mBP, mEHP, and mNP in three samples of pooled human milk have previously been reported by Calafat et al. ; mean concentrations were 1.1, 7.7, and 16.1 μg L−1, respectively. Thus, the levels of mBP and mEHP measured in this study were similar to previously reported values, although we found higher levels of mNP . mMP, mEP, and mBzP were present at the lowest levels, but were detectable, in almost all our samples. Thus the concentrations of these three phthalate monoesters were apparently generally higher in our milk samples than in previously tested pools of milk, in which they could not be detected .
The levels of phthalate monoesters observed in human milk reflect not only the level of exposure but the equilibrium between exposure, the metabolic clearance rate, and delivery to the milk, which in turn depends on the properties of the individual monoester. Thus, whereas phthalate monoesters with high water solubility, e.g. mEP, are found in the highest concentrations in urine , the more lipophilic monoesters mEHP and mNP were present in the highest concentrations in milk, which has a fat content of 2–4%. Furthermore, more lipophilic phthalate monoesters are present in urine and serum predominantly in the more hydrophilic glucoronidated forms [18, 27], whereas the concentrations measured in milk samples after enzymatic deglucoronidation, to determine the total (free and glucuronidated) content, showed that all the phthalate monoesters investigated in human milk were present mainly in the free form. This finding is in agreement with results reported elsewhere . The different distribution of individual monoesters and glucoronidated forms in urine, serum, and milk presumably reflects differences in water solubility.
Human exposure to phthalate diesters has been calculated using the concentrations of the metabolites in urine [35, 36] resulting in estimates of median daily intake of DEHP varying from 0.71 μg kg−1 day−1  to 13.8 μg kg−1 day−1 . Other estimates of exposure from food and the environment have shown important intake from food [4, 37] corresponding to 5–20 μg kg−1 bw day−1 . The data we present here on the concentration of phthalate monoesters in human milk cannot be used to estimate maternal intake of phthalate diesters because the relationship between the diester intake and the amount of monoesters delivered to the milk is unknown. However, results from this present study and that by Calafat et al.  show that phthalates can be incorporated into human milk and transferred to the nursing child.
Only mBP and mEHP could be detected in samples of consumer milk (from cows) and infant formula. The concentrations were at the same levels in consumer milk and infant formula. This was expected, because all the formulas tested were based on cows’ milk. There was also no difference between products from conventional farming and organic farming. Comparison with human milk seems to imply that exposure to most phthalates is lower for cows than for humans.
In this study we measured the metabolic monoesters rather than the parent phthalate diesters, because in animal studies specific monoesters have been shown to be the active principle for induction of reproductive and developmental toxicity [6, 12] and our aim was to obtain an estimate of potential exposure of infants via milk after maternal exposure. By analysing the monoesters we also avoided the common analytical problems of contamination with the parent phthalates, which are omnipresent in the environment. Metabolism of contaminating phthalate diesters can occur ex vivo in milk because of the presence of metabolising enzymes. We checked whether possible contamination with phthalate diesters during handling of the samples in the laboratory was likely to affect measured phthalate monoester levels by incubating milk samples without and with addition of acid to stop enzyme activity. The phthalate monoester levels were the same irrespective of whether or not acid had been added to the sample, indicating either that no contamination with phthalate diesters occurred during handling in the laboratory, or if it did occur, no degradation to the monoester occurred.
In this study it was not feasible to add acid to the milk immediately after sampling, because the milk was sampled in the homes of the participating women in multiple aliquots. Because the samples were stored at −20°C, however, enzyme activity was assumed to be minimal during storage. We cannot exclude the possibility that some contamination by diesters occurred during sampling, e.g. from phthalate-containing cosmetics on the mother’s skin or the use of breast pumps, which may be manufactured from phthalate-containing plastic parts. We also cannot exclude the possibility that some metabolism of contaminating phthalate diesters to monoesters occurred during the short time between sampling and freezing of the milk. Our observation of somewhat higher mEP and mBP levels in milk sampled by use of breast pumps suggests this could occur, although this remains to be verified. Our main interest has, however, been to estimate exposure of the infant via human milk, and any phthalate which ends up in the milk will cause exposure of the child who is fed with the milk. This is irrespective of whether the source of monoester contamination is the milk, per se, as it comes from the mother or because of contamination occurring during normal breast feeding (e.g. via mother’s skin) or sampling procedures (e.g. breast pump) for later feeding to the child.
In conclusion, the reported method is suitable for measurement of six phthalate monoesters in human milk with high sensitivity and accuracy. Levels of phthalate monoesters differ with levels of mBP, mEHP and mNP being highest. Large inter-individual variations were observed; for a few samples concentrations were high. Milk is the major source of nutrition for infants but it remains to be seen whether the phthalate monoester levels measured in human milk are a health hazard to infants.
The authors gratefully acknowledge financial support from Velux Fonden for instrumental equipment and financial support from the European Union, project Expored (QLK4-CT-2001-00269), and The Danish Medical Research Council (9700833, 9700909). We also thank John W. Brock PhD for his advice during the course of this project.