Extraction and Liquid Chromatography–Tandem Mass Spectrometry

Nov 27, 2016 - Extraction and Liquid Chromatography–Tandem Mass Spectrometry Detection of 3-Monochloropropanediol Esters and Glycidyl Esters in Infa...
2 downloads 24 Views 1MB Size
Article pubs.acs.org/JAFC

Extraction and Liquid Chromatography−Tandem Mass Spectrometry Detection of 3‑Monochloropropanediol Esters and Glycidyl Esters in Infant Formula Jessica K. Leigh* and Shaun MacMahon Center for Food Safety and Applied Nutrition, United States Food and Drug Administration, 5001 Campus Drive, College Park, Maryland 20740, United States ABSTRACT: A method was developed for the extraction of fatty acid esters of 3-chloro-1,2-propanediol (3-MCPD) and glycidol from infant formula, followed by quantitative analysis of the extracts using liquid chromatography−tandem mass spectrometry (LC-MS/MS). These process-induced chemical contaminants are found in refined vegetable oils, and studies have shown that they are potentially carcinogenic and/or genotoxic, making their presence in edible oils (and processed foods containing these oils) a potential health risk. The extraction procedure involves a liquid−liquid extraction, where powdered infant formula is dissolved in water and extracted with ethyl acetate. Following shaking, centrifugation, and drying of the organic phase, the resulting fat extract is cleaned-up using solid-phase extraction and analyzed by LC-MS/MS. Method performance was confirmed by verifying the percent recovery of each 3-MCPD and glycidyl ester in a homemade powdered infant formula reference material. Average ester recoveries in the reference material ranged from 84.9 to 109.0% (0.6−9.5% RSD). The method was also validated by fortifying three varieties of commercial infant formulas with a 3-MCPD and glycidyl ester solution. Average recoveries of the esters across all concentrations and varieties of infant formula ranged from 88.7 to 107.5% (1.0−9.5% RSD). Based on the validation results, this method is suitable for producing 3-MCPD and glycidyl ester occurrence data in all commercially available varieties of infant formula. KEYWORDS: 3-monochloropropanediol, 3-MCPD, 3-MCPD esters, glycidol, glycidyl esters, processing contaminants, infant formula, fat extraction



INTRODUCTION Vegetable oils commonly undergo industrial processing to eliminate unwanted tastes, colors, odors, and to remove components that could negatively impact shelf stability and nutritional value.1 However, industrial processing, particularly the application of high temperatures during deodorization, can result in the formation of undesirable process-induced chemical contaminants. In particular, fatty acid esters of 3-chloro-1,2propanediol (3-MCPD), 2-chloro-1,3-propanediol (2-MCPD), and glycidol have been found in a wide range of refined edible oils.2−8 Studies have suggested the formation of 2- and 3MCPD esters are the result of chemical reactions of triacylglycerol and partial acylglycerol precursors with reactive chlorine donors. Glycidyl esters arise from the radical rearrangement of diacylglycerols. Both mechanisms require the high temperatures (≥200 °C) used during deodorization.9−12 Because a number of recent studies suggest that MCPD and glycidyl esters may cause adverse toxicological effects,13−19 their presence in edible oils (and processed foods containing these oils) has been considered a potential health risk. A number of toxicological studies have shown that once ingested, fatty acid esters of MCPD and glycidol, or “bound” forms, are hydrolyzed with a high degree of efficiency, releasing the “free” forms of MCPD and glycidol.16,17,20 Therefore, numerous toxicological studies have been conducted to assess the health risks of bound 3-MCPD and glycidol. The European Scientific Committee on Food has labeled 3-MCPD as a nongenotoxic threshold carcinogen,21 and the Joint FAO/WHO This article not subject to U.S. Copyright. Published XXXX by the American Chemical Society

Expert Committee on Food Additives (JECFA) has established a maximum tolerable daily intake (TDI) of 2 μg/kg body weight daily of 3-MCPD.22 More recently, the European Food Safety Authority (EFSA) released an opinion that derived a TDI for 3-MCPD of 0.8 μg/kg body weight daily.23 The toxicological effects of 3-MCPD have been studied extensively, and research suggests the main effects of toxicity are focused on the kidneys and reproductive organs.13,24 Glycidol, on the other hand, has been classified as a genotoxic carcinogen by the International Agency for Research on Cancer (IARC), acting as a DNA alkylating agent.25 For this reason, it has been recommended that glycidol intake should be as low as reasonably achievable (ALARA).23 Studies are limited on the toxicity of 2-MCPD; however, research does suggest that, although structurally similar to 3-MCPD, the observed toxicological effects are different with the main effects of exposure observed in striated muscle tissues and the heart.14 Early analytical methodology for detecting and quantifying MCPD and glycidyl esters in edible oils was exclusively indirect analysis, which required a transesterification of the individual esters to form free MCPD and glycidol, followed by derivatization of the alcohols for analysis by GC-MS. These methods were relatively easy to carry out as they only required three standards (3-MCPD, 2-MCPD, and glycidol); however, Received: Revised: Accepted: Published: A

September 29, 2016 November 22, 2016 November 27, 2016 November 27, 2016 DOI: 10.1021/acs.jafc.6b04361 J. Agric. Food Chem. XXXX, XXX, XXX−XXX

Journal of Agricultural and Food Chemistry

Article

Table 1. Target 3-MCPD Diesters/Monoesters and Glycidyl Esters in the Analysis of Commercial Infant Formula Manufacturers in the United States compound

abbreviation

3-MCPD Diesters 1,2-dilinolenoyl-3-chloropropanediol Ln-Ln linoleoyl-linolenoyl-3-chloropropanediol Li-Ln 1,2-dilinoleoyl-3-chloropropanediol Li-Li palmitoyl-linoleoyl-3-chloropropanediol Pa-Ln oleoyl-linolenoyl-3-chloropropanediol Ol-Ln palmitoyl-linoleoyl-3-chloropropanediol Pa-Li oleoyl-linoleoyl-3-chloropropanediol Ol-Li 1,2-bis-palmitoyl-3-chloropropanediol Pa-Pa palmitoyl-oleoyl-3-chloropropanediol Pa-Ol 1,2-dioleoyl-3-chloropropanediol Ol-Ol stearoyl-linoleoyl-3-chloropropanediol St-Li palmitoyl-stearoyl-3-chloropropanediol Pa-St oleoyl-stearoyl-3-chloropropanediol Ol-St 1,2-distearoyl-3-chloropropanediol St-St

CAS Registry Number

compound

74875-97-1 1612870-93-5 74875-96-0 1246833-14-6 1246833-83-9 1246833-87-3 1336935-03-5 51930-97-3 1363153-60-9 69161-73-5 1246833-46-4 1185060-41-6 1336935-05-7 72468-92-9

CAS Registry Number

sn-1 3-MCPD Monoesters 1-linolenoyl-3-chloropropanediol Ln 1-linoleoyl-2-chloropropanediol Li 1-oleoyl-3-chloropropanediol Ol 1-stearoyl-3-chloropropanediol St

74875-99-3 74875-98-2 10311-82-7 22094-20-8

sn-2 3-MCPD Monoesters 2-linoleoyl-3-chloropropanediol 2Li 2-palmitoyl-3-chloropropanediol 2Pa 2-oleoyl-3-chloropropanediol 2Ol

1470071-08-9 20618-92-2 915297-48-2

Deuterated (d5) 3-MCPD Monoesters (Internal Standards) 1-palmitoyl-3-chloropropanediol-d5 Pad5 1346599-60-7 1-linoleoyl-3-chloropropanediol-d5 Lid5 1-oleoyl-3-chloropropanediol-d5 Old5 1-stearoyl-3chloropropanediol-d5 Std5 1795785-84-0 2-palmitoyl-3-chloropropanediol-d5 2Pad5 1329614-79-0

Deuterated (d5) 3-MCPD Diesters (Internal Standards) 1,2-dilinolenoyl-3-chloropropanediol-d5 Ln-Lnd5 linoleoyl-linolenoyl-3Li-Lnd5 chloropropanediol-d5 1,2-dilinoleoyl-3-chloropropanediol-d5 Li-Lid5 oleoyl-linolenoyl-3-chloropropanediolOl-Lnd5 d5 palmitoyl-linoleoyl-3Pa-Lid5 1246833-66-8 chloropropanediol-d5 oleoyl-linoleoyl-3-chloropropanediol-d5 Ol-Lid5 1,2-dioleoyl-3-chloropropanediol-d5 Ol-Old5 1246933-00-0 stearoyl-linoleoyl-3-chloropropanediolSt-Lid5 1246833-48-6 d5 palmitoyl-stearoyl-3-chloropropanediolPa-Std5 d5 oleoyl-stearoyl-3-chloropropanediol-d5 Ol-Std5 1,2-distearoyl-3-chloropropanediol-d5 St-Std5 1246818-85-8 sn-1 3-MCPD Monoesters 1-lauroyl-3-chloropropanediol La 1-myristoyl-3-chloropropanediol My 1-palmitoyl-3-chloropropanediol Pa

abbreviation

20542-96-5 30557-03-0 30557-04-1

while recent indirect methods produce accurate results,7,26 some of the original indirect methods were shown to be inaccurate.27,28 Due to the potential inaccuracies of the early indirect methods, direct methods involving the analysis of the intact esters were developed.28−35 Because direct methodology involves the analysis of the intact esters as they occur in the edible oils, the possibility of the creation or destruction of MCPD or glycidyl esters prior to analysis is eliminated. In particular, MacMahon et al. published direct LC-MS/MS methods for the detection and quantitation of 2- and 3-MCPD monoesters, 2- and 3-MCPD diesters, and glycidyl esters in edible oils.34−36 These methods require relatively simple sample preparation and reach limits of detection that are sufficient for the quantitation of MCPD and glycidyl esters in oils and, if coupled with a suitable extraction method, processed foods, particularly infant formula. Recently, increased attention has focused on the use of refined vegetable oils as the primary fat source in commercial infant formulas. Due to infants’ low body weights and formula commonly being used as the sole source of infant nutrition, there is a need for methodology to detect MCPD and glycidyl

Glycidyl Esters La-GE Li-GE Ln-GE My-GE Ol-GE Pa-GE St-GE

glycidyl glycidyl glycidyl glycidyl glycidyl glycidyl glycidyl

laurate linoleate linolenate myristate oleate palmitate stearate

1984-77-6 24305-63-3 51554-07-5 7460-80-2 5431-33-4 7501-44-2 7460-84-6

glycidyl glycidyl glycidyl glycidyl glycidyl glycidyl glycidyl

Deuterated Glycidyl Esters (Internal Standards) laurate-d5 La-GEd5 1329563-35-0 linoleate-d5 Li-GEd5 1246834-15-0 linolenate-d5 Ln-GEd5 1287393-54-7 myristate-d5 My-GEd5 1330180-72-7 oleate-d5 Ol-GEd5 1426395-63-2 palmitate-d5 Pa-GEd5 1794941-80-2 stearate-d5 St-GEd5 1346598-19-3

esters in formula, to determine infant exposure, and to assess risk. Although there are a number of published analytical methods for analyzing these contaminants as they occur in oils, there are limited published methods for extracting and quantifying these contaminants in processed foods.37−40 Extraction methods for infant formula pose a particular challenge given the complexity of the matrix, consisting of dried, thoroughly homogenized fats, carbohydrates and proteins. The goal of this work was to develop and validate a simple yet rugged procedure for extracting MCPD and glycidyl esters found in the fat content of a variety of commercial infant formulas, and then analyze these infant formula fat extracts for their ester content utilizing the published direct methodology of MacMahon et al.34,35



MATERIALS AND METHODS

Reagents and Materials. Ethyl acetate (HPLC grade, EtOAC), methanol (Optima LC/MS, MeOH), acetonitrile (Optima LC/MS, ACN), ethyl ether anhydrous (BHT stabilized, certified ACS, Et2O), isopropanol (Optima LC/MS, IPA), water (Optima LC/MS, H2O), iso-octane (HPLC grade, ISO), acetone (HPLC grade, ACE), and B

DOI: 10.1021/acs.jafc.6b04361 J. Agric. Food Chem. XXXX, XXX, XXX−XXX

Journal of Agricultural and Food Chemistry

Article

sodium sulfate, anhydrous (granular, certified ACS, Na2SO4) were purchased from Fisher Scientific (Pittsburgh, PA, USA). Hexane (Chromasolv for HPLC), methyl tert-butyl ether (Chromasolv for HPLC, MTBE), and 1-butanol (HPLC, BuOH) were obtained from Sigma-Aldrich (St. Louis, MO, USA), and dichlormethane (HPLC grade, CH2Cl2) was manufactured by Acros Organics (Morris Plains, NJ, USA) and purchased from Fisher Scientific. UHP 300 Ultra-High Purity nitrogen (N2) was obtained from Airgas (Hyattsville, MD, USA) and used for drying organic solutions. Formic acid (98%, MS grade) and ammonium formate (99%, MS grade), used in the preparation of the LC mobile phase, were purchased from SigmaAldrich. Solid phase extraction (SPE) cartridges (Supelclean LC-Si 500 mg/3 mL, Supelclean LC-Si 1000 mg/6 mL, and Supelclean LC-18 1000 mg/6 mL), used in the clean-up procedure for the infant formula fat extracts, were manufactured by Supelco Inc. (Bellefonte, PA, USA) and purchased from Sigma-Aldrich. Generic brands of milk, heavy whipping cream, animal rennet, palm olein oil, sunflower oil, soybean oil, and coconut oil were purchased from local supermarkets or online suppliers, and were used in the formulation of the homemade infant formula reference material (described below). Infant bifidum (Bif idobacterium infantis), lactose, nutritional yeast flakes, bovine gelatin, and acerola powder, also used in the recipe for homemade infant formula, were purchased from the Radiant Life Company (www.radiantlifecatalog.com) as part of the Nourishing Traditions Kit for Homemade Baby Formula. All commercial infant formulas used in the method development were purchased from local supermarkets or online suppliers. All 3-MCPD diesters/monoesters, glycidyl esters, and deuterated internal standards (except for Pa-Std5) (listed in Table 1) were purchased from Toronto Research Chemicals (Toronto, ON, Canada) and used without further purification. Pa-Std5 was synthesized according to a literature procedure.41 Individual stock solutions of each compound were prepared in IPA and stored at −20 °C for up to one year. Working spike solutions of the 3-MCPD diesters, 3-MCPD monoesters and glycidyl esters, and all deuterated ester internal standards were prepared in IPA at concentrations of 10 μg/mL, 10 μg/ mL, and 5 μg/mL, respectively. These solutions were used to fortify infant formula samples for validation and to prepare standard curve samples and were stored at approximately 5 °C for up to 6 months. Preparation of Homemade Infant Formula. The preparation of the homemade infant formula reference material (RM) was adapted from an online recipe from the Weston A. Price Foundation (www. westonaprice.org/childrens-health/formula-homemade-baby-formula/ ). The ingredients and amounts used in the formulation of the infant formula can be found in Table 2.

were slowly poured through a cheese cloth draped over an empty beaker in order to collect the whey. The homemade infant formula reference material was prepared using the following procedure: The lactose, gelatin, and half of the water (310.5 g) were poured into a beaker and heated to approximately 40−50 °C for 15−20 min, or until the lactose and gelatin had dissolved. The mixture was then removed from the heat, and the remaining water and coconut oil were stirred into the mixture until the coconut oil had melted. The remaining ingredients and the water mixture were then poured into a blender and blended approximately 20 s to ensure homogeneity. The blended infant formula was then poured into separate beakers (so that no beaker was more than 1/4 to 1/2 full), covered (using a piece of filter paper and a rubber band), and placed in a −80 °C freezer until freeze-drying. The homemade infant formula samples were freeze-dried in a VirTis Genesis 25L Pilot lyophilizer (SP Scientific, Warminster, PA). The freeze-drying protocol consisted of cooling the sample chamber and condenser to −40 °C before placing the frozen (and covered) infant formula samples inside the chamber. After cooling, the vacuum was initiated with no minimum setting, but the pressure typically equilibrated in the range of 10−20 mTorr. The initial shelf temperature was set to −25 °C and incrementally increased to 25 °C over a period of 36 h (Table 3). After 36 h, the samples were held

Table 3. Temperature Steps during the Freeze-Drying Process of Homemade Infant Formula

amount (g)

% fat

ingredient

amount (g)

% fat

water milk liquid whey heavy cream lactose infant bifidum nutritional yeast

621.00 172.50 21.30 10.65 20.13 0.45 3.90

0 0 0 38 0 0 3

gelatin acerola palm olein oil sunflower oil soybean oil coconut oil

3.75 0.47 3.75 1.88 1.88 1.88

0 0 100 100 100 100

time (hours)

−25 −15 −5 0 10 25

0 2 4 6 24 36

at 25 °C for an additional 36 h before being removed from the freezedryer. The dried formula samples were transferred into a single beaker and mixed thoroughly to ensure the final sample was homogenized. The percent fat in the homemade formula was calculated by summing the grams of fat contributed by each ingredient (Table 2) divided by the final mass of the dried formula. Preparation of Infant Formula Extracts. A 2 g portion of dried, powdered infant formula was weighed in a tared Corning polypropylene 50 mL conical centrifuge tube (Corning Inc., Corning, NY, USA). A 12 mL portion of EtOAc and 12 mL H2O (LC/MS grade) were added to the tube, and the mixture was vortexed using a Fisher Analog vortex mixer (Fisher Scientific) for 10 s before being placed in a Thermo MaxQ 6000 incubated/refrigerated shaker (Fisher Scientific), set at 35 °C and 500 rpm, for 1.5 h. Following heated shaking, the sample was centrifuged at 14500g for 20 min using a Thermo Sorvall Legend XTR centrifuge equipped with a Thermo Fiberlite F13-14x50cy fixed-angle rotor (Fisher Scientific). A 10 g portion of Na2SO4 was then added to the sample, and the mixture was vortexed for 10 s before a second round of centrifugation (14500g, 20 min). The second centrifugation step produced a defined organic phase (top layer), which was removed and transferred to an empty, weighed 16 × 125 mm round-bottom glass centrifuge tube (Fisher Scientific) (taking care not to transfer the solid or water phases during pipetting) and dried under N2 using a Techne sample concentrator equipped with a Techne Dri-Block DB-3 heater (Bibby Scientific, Burlington, NJ, USA) set to 60 °C. The dried sample, which yielded an oily, yellow-orange extract, was set aside for subsequent extractions. An additional 12 mL EtOAc was added to the 50 mL centrifuge tube containing the infant formula, and the sample was vortexed for 10 s before being placed in the heated shaker (35 °C and 500 rpm) for 30 min. The sample was removed from the shaker and centrifuged at 14500g for 20 min, the organic layer was transferred to the glass tube containing the oil extract from the first extraction, and the solvent was evaporated. This EtOAc extraction procedure was repeated once more,

Table 2. Ingredients Used in the Preparation of Homemade Infant Formula ingredient

shelf temperature (°C)

Liquid whey was prepared by slowly heating a beaker of 240 g of milk to 32 °C, and then adding 30 g of a mixture of 2.5 g of animal rennet and 120 g DI water. (The temperature was monitored carefully as heating above 32 °C will denature the milk proteins.) The milk/ rennet/water solution was then removed from heat, stirred for 30 s, and allowed to sit at room temperature for approximately 1 h. After this time, the milk formed a curd that held a knife cut. The curd was cut inside the beaker in a criss-cross pattern, allowing the liquid whey to separate from the curd. The contents of the beaker (curd + whey) C

DOI: 10.1021/acs.jafc.6b04361 J. Agric. Food Chem. XXXX, XXX, XXX−XXX

Journal of Agricultural and Food Chemistry

Article

Table 4. Sciex 5500 QTRAP MS/MS Conditions Q1 mass (Da)

Q3 mass (Da)

compound ID

661.510 661.510 661.510 659.510 659.510 659.510 663.510 663.510 663.510 633.510 633.510 633.510 395.300 395.300 397.300 399.200 401.200 371.210 373.210 656.500 656.500 656.500 654.500 654.500 654.500 658.500 658.500 658.500 628.500 628.500 628.500 390.300 390.300 392.300 390.300 390.300 392.300 394.200 396.200 366.210 368.210

608.500 364.500 268.200 606.500 362.300 360.300 610.510 364.300 362.300 580.500 336.200 268.200 263.200 245.200 263.200 267.200 267.200 239.200 239.200 603.510 359.300 263.200 601.500 357.300 355.300 605.500 359.300 357.300 575.500 331.200 263.200 263.200 245.200 263.200 263.200 245.200 263.200 267.200 267.200 239.200 239.200

St-Lid5.1 St-Lid5.2 St-Lid5.3 Ol-Lid5.1 Ol-Lid5.2 Ol-Lid5.3 Ol-Std5.1 Ol-Std5.2 Ol-Std5.3 Pa-Lid5.1 Pa-Lid5.2 Pa-Lid5.3 Lid5.1 Lid5.2 Lid5.3 Std5.1 Std5.2 2Pad5.1 2Pad5.2 St-Li.1 St-Li.2 St-Li.3 Ol-Li.1 Ol-Li.2 Ol-Li.3 Ol-St.1 Ol-St.2 Ol-St.3 Pa-Li.1 Pa-Li.2 Pa-Li.3 Li.1 Li.2 Li.3 2Li.1 2Li.2 2Li.3 St.1 St.2 2Pa.1 2Pa.2

internal standard

DP

EP

CE

CXP

St-Lid5.1 St-Lid5.2 St-Lid5.3 Ol-Lid5.1 Ol-Lid5.2 Ol-Lid5.3 Ol-Std5.1 Ol-Std5.2 Ol-Std5.3 Pa-Lid5.1 Pa-Lid5.2 Pa-Lid5.3 Lid5.1 Lid5.2 Lid5.3 Lid5.1 Lid5.2 Lid5.3 Std5.1 Std5.2 2Pad5.1 2Pad5.2

85 85 85 85 85 85 125 85 85 85 85 85 65 65 65 75 75 65 65 85 85 85 85 85 85 125 85 85 85 85 85 65 65 65 65 65 65 75 75 65 65

4 4 4 4 4 4 10 4 4 8 4 4 4 4 4 4 4 10 10 4 4 4 4 4 4 10 4 4 8 4 4 4 4 4 4 4 4 4 4 10 10

21 28 28 19 31 21 23 26 26 20 25 25 16 16 16 16 16 16 16 21 28 28 19 31 21 23 26 26 20 25 25 16 16 16 16 16 16 16 16 16 16

14 8 16 12 8 8 7 8 8 7 16 12 14 14 14 7 7 12 12 14 8 16 12 8 8 7 8 8 7 16 12 14 14 14 14 14 14 7 7 12 12

and the mass of the final dried oil extract was calculated by subtracting the mass of the empty glass tube from the mass of the extract + tube. Ready-to-feed and liquid concentrate formulas were also analyzed according to the described procedure with the exception that the amount of formula to be extracted was calculated based on the volume of liquid that was equivalent to 2 g of powdered infant formula. Sample Clean-Up of Infant Formula Extracts. The infant formula extract solution was prepared by pipetting 250 μL of 5 μg/mL deuterated internal standard spiking solution/g of oil extract into the glass centrifuge tube containing the oil extract. A 2 mL portion of 80:20 EtOAc/MTBE (v/v) was added to the tube, the solution was vortexed for 10 s, and the contents were transferred to a 5 mL volumetric flask. A 1 mL portion of 80:20 EtOAc/MTBE was then added to the glass tube, vortexed for 10 s, and again transferred to the 5 mL volumetric flask. This procedure was repeated once more, totaling 4 mL, and the flask was then filled to the 5 mL mark with 80:20 EtOAc/MTBE. An aliquot of the oil extract solution equivalent to 20 mg of extract was pipetted into an empty glass centrifuge tube and the target analytes were separated from the matrix using solid-phase extraction

(SPE) according to a previously published procedure for the analysis of 3-MCPD diesters.35 In brief, the solvent was evaporated from the aliquoted oil extract, reconstituted in 2 mL 2:98 Et2O/hexane (v/v), sonicated for 15 s, and added to a 1000 mg/6 mL Si SPE cartridge that was preconditioned with 5 mL MeOH, 5.5 mL CH2Cl2, and two 6 mL portions of 2:98 Et2O/hexane. The sample was collected in a new glass centrifuge tube. Three more 2 mL portions of 2:98 Et2O/hexane were added to the original glass tube containing the oil extract, sonicated for 15 s, and added to the SPE cartridge. A final volume of 5.5 mL of 2:98 Et2O/hexane was added directly to the SPE cartridge resulting in a total elution volume of 13.5 mL. The solvent from the collected sample was evaporated, the dried extract was reconstituted in 0.5 mL IPA, and the sample was transferred to an HPLC vial for analysis. A second aliquot (20 mg extract) of the oil extract solution was pipetted into a 1000 mg/6 mL C18 SPE cartridge for clean-up according to a published procedure for the analysis of 3-MCPD monoesters and glycidyl esters.34 In brief, the C18 SPE cartridge was preconditioned with 6 mL ACN and an aliquot of the oil extract solution (equivalent to 20 mg of extract) was pipetted directly into the SPE cartridge. A 14 mL volume of ACN was added to the cartridge, D

DOI: 10.1021/acs.jafc.6b04361 J. Agric. Food Chem. XXXX, XXX, XXX−XXX

Journal of Agricultural and Food Chemistry

Article

and the sample was collected in a new glass centrifuge tube. The solvent was evaporated from the collected sample, and the dried extract residue was reconstituted in 2 mL of 20:80 EtOAc/hexane (v/ v). The reconstituted sample was sonicated for 15 s and added to a 500 mg/3 mL Si SPE cartridge that was preconditioned with 3 mL 20:80 EtOAc/hexane. The sample was collected in a new glass centrifuge tube. Three more 2 mL portions of 20:80 EtOAC/hexane were added to the glass tube that contained the oil extract residue, sonicated for 15 s, and added to the Si SPE cartridge. A final volume of 1 mL of 20:80 EtOAC/hexane was added directly to the SPE cartridge resulting in a total elution volume of 9 mL. The solvent from the collected sample was evaporated, and the monoester/glycidyl ester method was slightly modified by reconstituting the final dried solution in 0.5 mL (instead of 1 mL) of IPA to maintain the analytical concentration of the original method, which was based on a clean-up of 40 mg of oil. The reconstituted sample was transferred to an HPLC vial for analysis. All SPE clean-ups were conducted using a Supelco Visiprep SPE vacuum manifold (Sigma-Aldrich) and samples were evaporated under N2 using a Techne sample concentrator equipped with a Techne DriBlock DB-3 heater. Dried samples were reconstituted in IPA and vortexed using a Fisher Analog vortex mixer followed by sonication in a Branson 2510 ultrasonic cleaner (Sigma-Aldrich). Samples were transferred to 2 mL Supelco clear glass autosampler vials (12 × 32 mm) with Supelco pre-slit PTFE/silicone septum polypropylene screw caps (Sigma-Aldrich) for analysis by LC-MS/MS. Instrumental Analysis. Separation of the analytes was performed using a Shimadzu Prominence UFLC XR liquid chromatography system (Shimadzu, Columbia, MD, USA) equipped with an Agilent Pursuit XRs C18 2.0 × 150 mm, 3.0 μm particle size analytical column (Agilent, Santa Clara, CA, USA). The column was held at 30 °C. The diester and glycidyl esters were separated using the following mobile phases: 2 mM ammonium formate/0.05% formic acid in 92:8 MeOH:H2O (mobile phase A) and 2 mM ammonium formate/ 0.05% formic acid in 98:2 IPA:H2O (mobile phase B). Monoesters were separated using 2 mM ammonium formate/0.05% formic acid in 75:25 MeOH:H2O (mobile phase A) and 2 mM ammonium formate/ 0.05% formic acid in 98:2 IPA:H2O (mobile phase B). A Sciex 5500 QTRAP mass spectrometer equipped with an electrospray ionization (ESI) source (Sciex, Foster City, CA, USA) in positive ion mode was used for MS/MS analysis of the analytes, and Analyst 1.6.2 software (Sciex) was used to control the LC and MS systems. Samples were analyzed for their diester, monoester, and glycidyl ester content according to previously validated LC-MS/MS procedures.34,35 MS/MS data were collected in scheduled multiple reaction monitoring (MRM) mode as previously reported34,35 with the addition of seven other deuterated diester/monoester standards (St-Lid5, Ol-Lid5, Ol-Std5, Pa-Lid5, Lid5, Std5, and 2Pad5) that have recently become commercially available (MRM conditions for the new standards are listed in Table 4). Quantitation. A nine point calibration curve was used for diester, monoester, and glycidyl ester quantitation. Calibration sample solutions were prepared in IPA in concentrations of 1, 2, 5, 10, 25, 50, 125, 250, and 400 ng/mL of each ester, along with an internal standard concentration of 50 ng/mL. The ratio of the chromatographic peak area to that of the corresponding internal standard was used with a weighting of 1/X2 to ensure correct curve fit at lower concentrations. Linear curves were used for most standards. Quadratic curves were used for symmetrical 3-MCPD diesters. Fit correlation was R2 ≥ 0.990.

Figure 1. Fat recovery from commercial premium, soy, and gentle infant formulas using various organic solvents for fat extraction.

Figure 2. Fat recovery from commercial premium, soy, and gentle infant formulas using various extraction parameters.

several studies have shown that similar conditions can potentially cause the interconversion between and/or destruction of MCPD and glycidyl esters.28,45 Therefore, not only was it necessary to develop a procedure that could facilitate the extraction of these fatty acid esters from such a complex matrix, but the method also needed to avoid the use of any reagents or conditions that could lead to a chemical reaction with the analyte(s) of interest. Finally, the procedure needed to be suitable for extracting the fat in such a way that it could be analyzed using the established LC-MS/MS method of MacMahon et al.34,35 Method Development. In general, initial extraction experiments consisted of a basic liquid−liquid extraction where an organic solvent was added to an amount of powdered infant formula dissolved in water, and the mixture was shaken to facilitate the partitioning of the fat into the organic phase. The sample was centrifuged to enhance and accelerate the separation of the organic and aqueous phases, followed by a transfer of the organic phase to an empty vial so that the sample could be dried, yielding an oily fat extract. The extraction



RESULTS AND DISCUSSION For extraction purposes, infant formula is a challenging food matrix given the complexity of its components, including lipids/ fats, carbohydrates, proteins, vitamins, and minerals in thoroughly homogenized, dried form. Traditional methods of fat extraction from foods such as baked goods and dairy products typically require sample pretreatment involving a harsh hydrochloric acid hydrolysis procedure.42−44 However, E

DOI: 10.1021/acs.jafc.6b04361 J. Agric. Food Chem. XXXX, XXX, XXX−XXX

Journal of Agricultural and Food Chemistry

Article

Table 5. Recoveries of Individual Esters (Present above the LOQ) in Homemade Infant Formula Using Various Extraction Solvents/Parameters average percent recovery, percent RSD (where applicable)a analyte

theor conc (ng/g)

A (n = 1)

B (n = 1)

C (n = 1)

D (n = 1)

E (n = 3)

F (n = 3)

G (n = 3)

151.5 107.1 98.6 99.1 118.1

96.2, 83.6, 86.7, 101.4, 98.7,

10.2 9.5 6.8 3.1 3.5

99.9, 82.4, 86.7, 82.1, 97.1,

3.8 3.6 6.5 3.1 4.4

107.5, 93.4, 90.7, 84.9, 95.3,

3.0 3.2 3.5 0.6 3.2

3.55 16.99 5.97 4.73 3.31

123.0 103.0 92.1 87.2 132.6

118.2 97.1 86.2 83.0 177.5

Monoesters 94.5 85.4 79.5 86.4 119.3

La-GE Li-GE Ln-GE My-GE Ol-GE Pa-GE St-GE

3.97 49.19 5.24 5.03 260.3 82.14 14.07

83.8 97.2 93.3 62.8 99.5 87.4 108.1

88.1 104.9 90.3 70.1 103.0 102.5 108.1

Glycidyl Esters 97.9 101.0 93.9 88.2 97.2 106.9 98.8

95.9 109.2 96.4 91.6 101.0 114.3 103.1

112.7, 106.1, 97.1, 93.2, 103.4, 106.7, 96.0,

2.0 6.7 5.7 5.8 3.3 5.0 6.0

99.2, 93.7, 92.1, 95.4, 91.7, 82.3, 92.7,

2.5 1.1 5.8 2.2 1.0 2.0 1.2

104.3, 95.7, 95.8, 97.8, 96.7, 85.8, 94.8,

3.3 6.5 9.5 7.7 4.4 3.0 7.1

Li-Li Pa-Li Ol-Li Pa-Pa Pa-Ol Ol-Ol St-Li Pa-St Ol-St Ol-Ln

5.23 25.40 32.91 20.55 120.0 67.66 9.39 5.60 11.82 4.86

97.6 114.5 118.5 118.8 90.0 81.1 91.7 119.0 99.0 99.8

100.7 112.6 110.6 118.8 90.0 85.7 91.7 130.8 97.3 99.4

Diesters 91.7 102.4 104.2 117.3 90.8 91.6 93.8 113.0 104.9 100.8

98.2 111.0 123.7 130.9 95.8 76.9 96.7 109.9 95.6 107.4

93.4, 116.9, 124.0, 133.8, 95.0, 71.2, 98.3, 108.7, 96.4, 91.4,

6.7 9.0 3.8 5.7 2.6 5.1 5.9 4.8 2.1 5.3

99.1, 95.5, 89.5, 106.1, 91.1, 87.9, 90.6, 94.4, 100.1, 98.2,

2.3 3.5 1.5 4.5 2.1 1.8 6.6 9.5 0.5 5.2

101.1, 94.9, 93.1, 109.0, 98.9, 91.6, 91.3, 95.5, 99.8, 99.3,

4.7 4.9 4.1 3.5 9.5 3.5 3.3 8.7 3.7 3.5

1Li 1Ol 1Pa 2Ol 2Pa

a

Extractions performed under the following conditions unless otherwise noted: centrifuge 14500g, no heated shaking, no Na2SO4. Column conditions: (A) EtOAc/MTBE/80:20 EtOAc-MTBE, (B) 80:20 EtOAc-MTBE, (C) EtOAc, (D) EtOAc with aqueous 4 M NaNO3, (E) 80:20 ISOACE, (F) EtOAc with Na2SO4, and (G) EtOAc with heated shaking (35 °C) and Na2SO4.

gastroesophageal reflux, etc.) that generally consist of hydrolyzed (or broken-down) milk proteins and/or additional ingredients for specific needs. It has been shown that the surface-active properties of milk proteins can lower the surface tension between the aqueous and organic layers, thereby resulting in emulsions where the interface of the phases cannot be distinguished.40 Consistent with the literature, Figure 1 shows the fat from soy-based formulas, which do not contain milk proteins, is effectively extracted from the matrix (fat recoveries of ∼90%) regardless of the organic extraction solvent. However, the extent of fat extraction from the premium and gentle formulas was strongly dependent on the organic solvent used (fat recoveries ranging from 20 to 80%), thus demonstrating the importance of developing an extraction procedure that would be widely applicable to all varieties of commercial infant formulas. Figure 1 also reveals the impact of solvent selection on fat recovery. A number of organic solvents with a wide range of polarities were investigated for fat extraction. Due to the range of polarities of the analytes of interest (i.e., relatively nonpolar MCPD diesters and moderately polar MCPD monoesters and glycidyl esters) it was important to select a solvent that ensured that all the target analytes of varying degrees of polarities were recovered. The percent of fat recovered from the soy formulas was relatively independent of solvent polarity. However, as shown in Figure 1, solvent polarity had a significant impact on the percent of fat extracted from premium and gentle infant

procedure was repeated twice more to ensure optimal fat extraction from the matrix. In order to assess the efficiency of the fat extraction during method development the mass of the fat recovered after each extraction was weighed and compared to the expected amount of fat from each sample, which was calculated using the nutrition labels on each infant formula sample (usually 27−30% fat). Initial method development focused on identifying the optimal volumes of aqueous and organic solvents to be added to 2 g of formula. A ratio of approximately 6 times the mass of sample used (12 mL organic + 12 mL H2O) was most effective in avoiding the formation of significant emulsions that could not be minimized/eliminated with centrifugation. In addition, a ratio of aqueous to organic solvent of at least 1:1 was also necessary, as experiments showed that a decreased water to organic ratio also resulted in unmanageable emulsions. Initial experiments during method development showed that fat extraction yields from various types of infant formulas differed drastically from one type of formula to the next when using the same set of extraction conditions. Commercial infant formulas in the United States for the most part can be categorized into three basic varieties: (1) premium infant formulas, which consist of fully intact milk proteins, (2) soy infant formulas, which do not contain any milk proteins and are replaced with soy proteins, and (3) gentle infant formulas, which encompass a variety of formulas for special needs (e.g., complications with fat digestion, malabsorption syndromes, F

DOI: 10.1021/acs.jafc.6b04361 J. Agric. Food Chem. XXXX, XXX, XXX−XXX

St-GE.1

formulas, with increasing solvent polarity (highest yields with polar organic solvents EtOAc and BuOH) maximizing fat extraction yields. Due to the relatively hydrophobic character of the milk proteins in the infant formula, one possible explanation is that increasing the polarity of the organic solvent minimizes interactions between the milk proteins and the solvent, allowing for a more distinct separation between the organic and aqueous phases. In addition, it is possible that more polar solvent may allow these milk proteins to precipitate, which may explain the presence of a small solid layer between the organic and aqueous phases, which was observed after centrifugation. Chlorinated solvents were not investigated as potential extraction solvents to avoid any possible addition of Cl− ion that could react with the analytes of interest. As shown in Figure 1, BuOH appeared to be an ideal candidate for extraction solvent, with the greatest fat recoveries observed using this solvent across all varieties of infant formula. However, because it is relatively non-volatile, BuOH was difficult to evaporate. Therefore, EtOAc was determined to be the more suitable extraction solvent as it is more volatile and allowed for faster evaporation. Although it was evident that a relatively more polar extraction solvent was an ideal choice for increasing fat extraction, it was apparent that solvent selection alone was not enough to obtain ≥90% fat extraction from all varieties of infant formula. It was necessary to optimize other critical extraction parameters in addition to solvent selection in order to maximize fat extraction across all varieties of infant formula. Using EtOAc as the extraction solvent, a number of parameters, including extended shaking, sonication, elevated centrifugation speeds, and salt addition were investigated. As shown in Figure 2, centrifugation at 14500g and addition of Na2SO4 to the samples following centrifugation at lower speeds (∼3000g) significantly enhanced fat extraction for all formula varieties. Higher centrifugation speeds not only allowed for more rapid separation of the aqueous and organic phases, but reduced or eliminated emulsions, resulting in enhanced separation of the organic/aqueous layers. Sodium sulfate (Na2SO4) was also effective at minimizing emulsions, and combining both Na2SO4 and higher centrifugation speeds minimized emulsions considerably, even when less polar extraction solvents (that had originally resulted in unmanageable emulsions) were used. In addition, in comparison to the other extraction parameters that produced acceptable fat recoveries (≥90%), use of both Na2SO4 and centrifugation at 14500g resulted in the highest reproducibility between extraction samples (Figure 2). Method Performance: Homemade Infant Formula. Although the previously described extraction conditions effectively recovered ≥90% of the expected fat in infant formula samples, it was not clear whether the individual MCPD monoesters/diesters and glycidyl esters would be effectively recovered from the formula samples. Because of their varying degrees of polarity (MCPD diesters are less polar than triglycerides, monoesters and glycidyl esters are more polar) there was a possibility that the chosen extraction solvent and/or extraction parameters may not extract all esters with the same degree of efficiency. Therefore, it was necessary to determine how the extraction solvent/parameters influenced the extraction efficiency of each individual ester. Traditional methods of confirming method performance typically involve spiking the sample matrix with a known concentration of the analytes of interest. However, particularly in solid samples, this method is not highly effective for verifying extraction efficiency because

100.3, 4.1 102.3, 3.8 99.2, 1.8

105.6, 7.9 103.3, 3.6 102.5, 1.7

100.7, 4.5 100.2, 3.3 100.2, 2.3

Ol-GE.1 Li-GE.1

98.4, 5.1 96.8, 6.9 96.4, 2.5

Ln-GE.1

25 250 2000

97.7, 4.2 101.8, 4.4 99.8, 2.7 25 250 2000

101.5, 2.9 97.1, 5.9 95.8, 3.0 98.6, 4.0 98.1, 4.4 100.8, 1.0

My-GE.1

106.9, 3.4 102.9, 4.1 104.1, 1.9

La-GE.1

99.4, 3.9 92.9, 4.7 93.1, 2.3

spike (ng/g infant formula)

Pa-GE.1

percent recovery, percent RSD Glycidyl Esters

St.1 My.1

Article

98.1, 8.0 95.2, 4.6 93.8, 4.1 102.7, 3.1 97.5, 2.19 98.3, 6.1

Ln.1 La.1

92.3, 5.0 93.3, 4.2 94.2, 3.2 101.5, 5.5 100.6, 4.4 98.9, 2.0

2Pa.1 2Ol.1

100.7, 4.1 94.5, 5.9 99.4, 3.5 100.8, 3.0 100.5, 7.8 96.9, 4.6 101.0, 5.2 94.8, 1.5 93.1, 3.6

percent recovery, percent RSD

2Li.1 1 Pa.1 1Ol.1 1Li.1 Monoesters

spike (ng/g infant formula)

99.3, 3.7 100.0, 3.7 98.6, 1.5

99.0, 6.0 94.5, 4.8 98.6, 5.4 100.8, 5.1 96.5, 3.1 95.8, 3.0 101.1, 5.1 101.9, 2.5 103.7, 1.1 97.5, 4.1 88.7, 1.7 96.7, 3.3 104.8, 9.5 94.6, 2.4 94.8, 5.3 100.7, 6.7 100.5, 3.5 99.4, 6.6 99.8, 3.9 98.6, 2.6 99.1, 3.6 98.7, 2.1 93.2, 3.3 95.1, 4.1 102.0, 5.0 100.5, 7.6 100.6, 5.8 100.3, 2.4 95.9, 4.6 98.2, 2.6 98.4, 5.2 100.2, 6.6 107.5, 5.6 100.7, 1.8 97.7, 1.2 98.0, 1.9 99.2, 3.5 94.8, 1.8 95.2, 3.2 25 250 2000

95.8, 5.7 97.8, 1.4 98.4, 3.4

St-St St-Li Pa-St Pa-Pa Pa-Ol Pa-Ln Pa-Li Ol-St

percent recovery, percent RSD

Ol-Ol Ol-Ln Ol-Li Ln-Ln Li-Ln Li-Li Diesters

spike (ng/g infant formula)

Table 6. Average Method Performance for 3-MCPD Diesters, 3-MCPD Monoesters, and Glycidyl Esters (n = 6 at Each Concentration, with Duplicate Spikes Run on Different Days)

Journal of Agricultural and Food Chemistry

G

DOI: 10.1021/acs.jafc.6b04361 J. Agric. Food Chem. XXXX, XXX, XXX−XXX

Journal of Agricultural and Food Chemistry

Article

also validated using a traditional procedure with spiked samples. It was essential to validate the performance of the method across all infant formula varieties. For this reason, three different commercial infant formulas, a premium, soy, and gentle infant formula variety, were chosen for method validation. To generate validation data, four 2 g samples of each infant formula variety (a total of 12 samples) were fortified, generating infant formula samples that were “blank”, 25 ng/g, 250 ng/g, and 2000 ng/g, respectively, of each ester standard. While the RM confirms the method is suitable for extracting contaminants incorporated into infant formula, these spiking experiments verified that the method performed adequately over the desired dynamic range and did not experience matrix effects/interferences from any of the common infant formula varieties. Although no spiking solution was added to the blank samples, many of the esters were already present in the commercial infant formulas. Concentrations of bound 3-MCPD and bound glycidol were 0.71 and 0.11 μg/g, respectively, for the blank premium formula, 0.35 and 0.051 μg/g, respectively, for the blank soy formula, and 0.66 and 0.18 μg/g, respectively, for the blank gentle formula. Therefore, it was necessary to subtract the concentrations of each ester found in the blank samples from the total ester concentrations in each fortified sample in order to determine recovery of the spiked esters. The internal standard spiking solution was added to the samples following the extraction but prior to SPE clean-up since it was expected that any solution applied to the surface of the solid infant formula would be fully recovered by the extraction. In addition, it was necessary to know the amount of fat that was extracted to determine the appropriate amount of internal standard solution. Internal standard solution was added to each fat extract based on 250 μL (5 μg/mL solution)/g of oil extract, generating sample extracts that were 1.25 μg/g of each ester internal standard. Because the masses of the sample extracts varied for each extraction, the final spiked 3-MCPD and glycidyl ester concentrations in the validation samples also differed for every sample. The final spiked sample concentrations (after sample clean-up) were calculated by multiplying the concentration of the esters in the 5 mL extract solution by the volume of the extract solution added to the SPE cartridge for sample clean-up, which equals the mass of the esters loaded onto the SPE cartridge. This mass was divided by the final volume (0.5 mL) of the reconstituted esters following sample clean-up. The final ester concentrations in the spiked validation samples were approximately 0, 3.7, 37, and 290 ng/g, respectively. The extracts of spiked samples were analyzed using previously published methodology for quantifying 3-MCPD diesters,35 3-MCPD monoesters, and glycidyl esters.34 Average recoveries for the 3-MCPD diesters and glycidyl esters in all three varieties of infant formulas ranged from 88.7 to 107.5% (relative standard deviation (RSD) range of 1.1−9.5%) and 93.1−106.9% (RSD range of 1.0−6.9%), respectively. Average recoveries for 8 of the 10 3-MCPD monoesters were in the range of 92.3−102.7% (RSD range of 1.5−8.0); however, results for the remaining two analytes, 2Pa (65.7%) and St (77.7%), showed significant matrix suppression. Since publication of the direct methodology for analyzing 3-MCPD monoesters, deuterated internal standards for 2Pa and St have become commercially available, and were therefore included in the method (Table 4) to correct recoveries for these analytes. After incorporation of these internal standards, recoveries across the three spiking concentrations for 2Pa and St

the analytes are more easily recovered when spiked on top of a matrix rather than being incorporated within the sample. Therefore, it was desirable to investigate the method performance beyond simple spiked samples. Since there was no commercially available infant formula reference material with known values for MCPD and glycidyl esters, an in-house homemade infant formula reference material (RM) was prepared using ingredients that are typically found in commercial infant formulas. Prior to preparation of the RM, four refined edible oils (palm, sunflower, soybean, and coconut), which encompass those typically found in commercial formulas, were analyzed using the direct methodology of MacMahon et al.34,35 to determine concentrations of each individual ester present in the oils. Concentrations of bound 3-MCPD and bound glycidol were 5.92 and 8.19 μg/g, respectively, for palm oil, 0.24 and 0.26 μg/g, respectively, for sunflower oil, 0.35 and 0.38 μg/g, respectively, for soybean oil, and 0.090 and 0.89 μg/g, respectively, for coconut oil. These oils were then used in the preparation of the RM so that expected concentrations of each ester in the final infant formula were known. The RM was prepared in liquid form and then freeze-dried, yielding a material that was visually and chemically similar to commercially available formulas. Because the homemade formula was prepared using milk, the sample was considered characteristic of a “premium” commercial formula, the most difficult infant formula matrix for fat extraction; therefore, method performance results using the RM were considered representative of all varieties of infant formula. A number of solvent combinations and parameters were considered for extracting the RM in order to ensure all of the esters of varying degrees of polarity were successfully recovered. In addition, differing solvent combinations were tested for each of the three extraction steps of the sample preparation (e.g., extraction 1, 100% EtOAc; extraction 2, 100% MTBE; extraction 3, 80:20 EtOAc/MTBE (v/v)) to determine if a combination of varying solvent polarities would enhance ester recoveries. Ester recoveries were also evaluated after addition of NaNO3 to the aqueous phase during the extraction to determine if increasing the ionic character of the aqueous phase would improve migration of lipids to the organic phase. The use of NaCl, commonly added to enhance liquid−liquid extraction separations, was avoided due to the potential for the Cl− to chemically transform the target analytes. Heated shaking of the samples at 35 °C was also considered for potentially enhancing the partitioning of the esters into the organic phase. As shown in Table 5, a number of the extraction solvent/ parameter combinations resulted in acceptable recoveries (85− 115%) for a majority of the esters. EtOAc, however, appeared to be the solvent of choice as shown by ester recoveries in the range of 79.5−117.3%, which were achieved using only EtOAc without modifying any additional extraction parameters (column C, Table 5). For this reason, a number of extraction parameters were investigated using ethyl acetate as the extraction solvent (columns E−G, Table 5). Optimal recoveries (ranging from 82 to 109%) were achieved with the addition of Na2SO4 and heated shaking at 35 °C (columns F and G, Table 5). Although heated and nonheated shaking of the samples (with the addition of Na2SO4) both produced acceptable ester recoveries, the data indicated that added heat yielded slightly improved recoveries, so it was ultimately used in the method validation. Method Performance: Spiked Samples. In addition to confirming method performance using a RM, the method was H

DOI: 10.1021/acs.jafc.6b04361 J. Agric. Food Chem. XXXX, XXX, XXX−XXX

Journal of Agricultural and Food Chemistry

Article

energy; CXP, collision cell exit potential; RM, homemade infant formula reference material

improved to 98.9−101.5% (RSD range of 2.0−5.5) and 102.5− 105.6% (RSD range of 1.7−7.9), respectively. Four additional deuterated diester internal standards were also added to the method (St-Lid5, Ol-Lid5, Ol-Std5, and Pa-Lid5) simply because they became commercially available before this study began; therefore, recoveries for St-Li, Ol-Li, Ol-St, and Pa-Li were not calculated using the internal standards described in previously published method protocol.35 The method recoveries and RSDs for all 3-MCPD/glycidyl esters across all infant formula varieties are reported in Table 6. The glycidyl ester and 3-MCPD monoester standards in the method were sufficient for the detection of all commonly consumed edible oils. However, there are currently no commercial sources of 3-MCPD diesters containing esters of lauric and/or myristic acid. Of the oils commonly used in infant formula production in the United States (coconut oil, corn oil, palm oil, palm olein, safflower oil, soybean oil, sunflower oil), only coconut oil contains significant quantities of lauric or myristic acid. While the portion of the 3-MCPD diesters in coconut oil containing lauric and myristic acid would not be detected, the method would detect the remaining 3-MCPD diesters, as well as all the 3-MCPD monoesters and glycidyl esters in the sample. In addition, published surveys of total bound 3-MCPD concentration in coconut oil indicate the concentrations are relatively low, with an average concentration below 200 ng/g.8 However, these diester standards will be added to the method if they become commercially available or will be synthesized using the published procedure.4 Applicability in Sample Analysis. The method presented for the extraction of MCPD and glycidyl esters in infant formula involves simple, liquid−liquid extraction procedures that do not involve the use of complex instrumentation or expensive equipment. In addition, this work is the first reported that has confirmed method performance beyond simple spiked samples by analyzing the recoveries of the intact esters in a homemade reference material. Based on ester recovery data from the reference material, as well as validation data obtained from a variety of spiked commercial infant formula samples, the performance of the extraction procedure is rugged and reliable, yielding highly reproducible results. Using the previously published LC-MS/MS methodology for detection and quantitation, this procedure is suitable for producing MCPD and glycidyl ester occurrence data in all varieties of commercially available infant formulas.





REFERENCES

(1) Pudel, F.; Benecke, P.; Fehling, P.; Freudenstein, A.; Matthäus, B.; Schwaf, A. On the necessity of edible oil refining and possible sources of 3-MCPD and glycidyl esters. Eur. J. Lipid Sci. Technol. 2011, 113, 368−373. (2) Hrnčiřík, K.; van Duijn, G. An initial study on the formation of 3MCPD esters during oil refining. Eur. J. Lipid Sci. Technol. 2011, 113, 374−379. (3) Matthäus, B.; Pudel, F.; Fehling, P.; Vosmann, K.; Freudenstein, A. Strategies for the reduction of 3-MCPD esters and related compounds in vegetable oils. Eur. J. Lipid Sci. Technol. 2011, 113, 380−386. (4) Destaillats, F.; Craft, B. D.; Dubois, M.; Nagy, K. Glycidyl esters in refined palm (Elaeis guineensis) oil and related fractions. Part I: Formation mechanism. Food Chem. 2012, 131, 1391−1398. (5) Destaillats, F.; Craft, B. D.; Sandoz, L.; Nagy, K. Formation mechanisms of monochloropropanediol (MCPD) fatty acid diesters in refined palm (Elaeis guineensis) oil and related fractions. Food Addit. Contam., Part A 2012, 29, 29−37. (6) Nagy, K.; Sandoz, L.; Craft, B.; Destaillats, F. Mass-defect filtering of isotope signatures to reveal the source of chlorinated palm oil contaminants. Food Addit. Contam., Part A 2011, 28, 1492−1500. (7) Kuhlmann, J. Determination of bound 2, 3-epoxy-1-propanol (glycidol) and bound monochloropropanediol (MCPD) in refined oils. Eur. J. Lipid Sci. Technol. 2011, 113, 335−344. (8) MacMahon, S.; Begley, T. H.; Diachenko, G. W. Occurrence of 3MCPD and glycidyl esters in edible oils in the United States. Food Addit. Contam., Part A 2013, 30, 2081−2092. (9) Svejkovská, B.; Doležal, M.; Velíšek, J. Formation and decomposition of 3-chloropropane-1, 2-diol esters in models simulating processed foods. Czech J. Food Sci. 2006, 24, 172. (10) Shimizu, M.; Vosmann, K.; Matthäus, B. Generation of 3monochloro-1, 2-propanediol and related materials from tri-, di-, and monoolein at deodorization temperature. Eur. J. Lipid Sci. Technol. 2012, 114, 1268−1273. (11) Freudenstein, A.; Weking, J.; Matthäus, B. Influence of precursors on the formation of 3-MCPD and glycidyl esters in a model oil under simulated deodorization conditions. Eur. J. Lipid Sci. Technol. 2013, 115, 286−294. (12) Ermacora, A.; Hrnčiřík, K. Influence of oil composition on the formation of fatty acid esters of 2-chloropropane-1, 3-diol (2-MCPD) and 3-chloropropane-1, 2-diol (3-MCPD) under conditions simulating oil refining. Food Chem. 2014, 161, 383−389. (13) Bakhiya, N.; Abraham, K.; Gürtler, R.; Appel, K. E.; Lampen, A. Toxicological assessment of 3-chloropropane-1, 2-diol and glycidol fatty acid esters in food. Mol. Nutr. Food Res. 2011, 55, 509−521. (14) Schilter, B.; Scholz, G.; Seefelder, W. Fatty acid esters of chloropropanols and related compounds in food: Toxicological aspects. Eur. J. Lipid Sci. Technol. 2011, 113, 309−313. (15) Buhrke, T.; Weißhaar, R.; Lampen, A. Absorption and metabolism of the food contaminant 3-chloro-1, 2-propanediol (3MCPD) and its fatty acid esters by human intestinal Caco-2 cells. Arch. Toxicol. 2011, 85, 1201−1208. (16) Abraham, K.; Appel, K. E.; Berger-Preiss, E.; Apel, E.; Gerling, S.; Mielke, H.; Creutzenberg, O.; Lampen, A. Relative oral bioavailability of 3-MCPD from 3-MCPD fatty acid esters in rats. Arch. Toxicol. 2013, 87, 649−659. (17) Appel, K. E.; Abraham, K.; Berger-Preiss, E.; Hansen, T.; Apel, E.; Schuchardt, S.; Vogt, C.; Bakhiya, N.; Creutzenberg, O.; Lampen, A. Relative oral bioavailability of glycidol from glycidyl fatty acid esters in rats. Arch. Toxicol. 2013, 87, 1649−1659. (18) Liu, M.; Gao, B.-Y.; Qin, F.; Wu, P.-P.; Shi, H.-M.; Luo, W.; Ma, A.-N.; Jiang, Y.-R.; Xu, X.-B.; Yu, L.-L. L. Acute oral toxicity of 3MCPD mono-and di-palmitic esters in Swiss mice and their cytotoxicity in NRK-52E rat kidney cells. Food Chem. Toxicol. 2012, 50, 3785−3791.

AUTHOR INFORMATION

Corresponding Author

*Phone: 240-402-3046. E-mail: [email protected]. ORCID

Jessica K. Leigh: 0000-0002-9348-4854 Notes

The authors declare no competing financial interest.



ABBREVIATIONS USED 3-MCPD, 3-monochloropropane-1,2-diol; LC-MS/MS, liquid chromatography−tandem mass spectrometry; SPE, solid-phase extraction; RSD, relative standard deviation; LOQ, limit of quantitation; IPA, isopropanol; EtOAc, ethyl acetate; MTBE, methyl tert-butyl ether; MeOH, methanol; CH2Cl2, dichloromethane; Et2O, ethyl ether; BuOH, butanol; ISO, iso-octane; ACE, acetone; Na2SO4, sodium sulfate; NaCl, sodium chloride; DP, declustering potential; EP, entrance potential; CE, collision I

DOI: 10.1021/acs.jafc.6b04361 J. Agric. Food Chem. XXXX, XXX, XXX−XXX

Journal of Agricultural and Food Chemistry

Article

(19) Liu, M.; Liu, J.; Wu, Y.; Gao, B.; Wu, P.; Shi, H.; Sun, X.; Huang, H.; Wang, T. T.; Yu, L. L. Preparation of five 3-MCPD fatty acid esters, and the effects of their chemical structures on acute oral toxicity in Swiss mice. J. Sci. Food Agric. 2016, DOI: 10.1002/jsfa.7805. (20) Seefelder, W.; Varga, N.; Studer, A.; Williamson, G.; Scanlan, F.; Stadler, R. Esters of 3-chloro-1, 2-propanediol (3-MCPD) in vegetable oils: significance in the formation of 3-MCPD. Food Addit. Contam., Part A 2008, 25, 391−400. (21) European Commission (EC). Opinion of the Scientific Committee on Food on 3-Monochloro-propane-1,2-diol (3-MCPD), Brussels, Belgium, 2001; http://ec.europa.eu/food/safety/docs/cs_ contaminants_catalogue_mcpd_out91_en.pdf (accessed Sept 2016). (22) World Health Organization (WHO). Food Additives Series: 48; WHO: Geneva, Switzerland, 2002; http://www.inchem.org/ documents/jecfa/jecmono/v48je18.htm (accessed Sept 2016). (23) EFSA. Panel on Contaminants in the Food Chain (CONTAM). Risks for human health related to the presence of 3- and 2monochloropropanediol (MCPD), and their fatty acid esters, and glycidyl fatty acid esters in food. EFSA J. 2016, 14. (24) Cho, W.-S.; Han, B. S.; Nam, K. T.; Park, K.; Choi, M.; Kim, S. H.; Jeong, J.; Jang, D. D. Carcinogenicity study of 3-monochloropropane-1, 2-diol in Sprague−Dawley rats. Food Chem. Toxicol. 2008, 46, 3172−3177. (25) World Health Organization (WHO), International Agency for Research on Cancer (IARC). IARC Monographs on the Evaluation of Carcinogenic Risks to Humans: Some Industrial Chemicals, Vol. 77, 2000; http://monographs.iarc.fr/ENG/Monographs/vol77/mono77. pdf (accessed Sept 2016). (26) Ermacora, A.; Hrnčiřík, K. A novel method for simultaneous monitoring of 2-MCPD, 3-MCPD and glycidyl esters in oils and fats. J. Am. Oil Chem. Soc. 2013, 90, 1−8. (27) Weißhaar, R. Determination of total 3-chloropropane-1, 2-diol (3-MCPD) in edible oils by cleavage of MCPD esters with sodium methoxide. Eur. J. Lipid Sci. Technol. 2008, 110, 183−186. (28) Haines, T. D.; Adlaf, K. J.; Pierceall, R. M.; Lee, I.; Venkitasubramanian, P.; Collison, M. W. Direct determination of MCPD fatty acid esters and glycidyl fatty acid esters in vegetable oils by LC−TOFMS. J. Am. Oil Chem. Soc. 2011, 88, 1−14. (29) Hori, K.; Koriyama, N.; Omori, H.; Kuriyama, M.; Arishima, T.; Tsumura, K. Simultaneous determination of 3-MCPD fatty acid esters and glycidol fatty acid esters in edible oils using liquid chromatography time-of-flight mass spectrometry. LWT-Food Sci. Technol. 2012, 48, 204−208. (30) Moravcova, E.; Vaclavik, L.; Lacina, O.; Hrbek, V.; Riddellova, K.; Hajslova, J. Novel approaches to analysis of 3-chloropropane-1, 2diol esters in vegetable oils. Anal. Bioanal. Chem. 2012, 402, 2871− 2883. (31) Dubois, M.; Tarres, A.; Goldmann, T.; Empl, A. M.; Donaubauer, A.; Seefelder, W. Comparison of indirect and direct quantification of esters of monochloropropanediol in vegetable oil. J. Chromatogr. A 2012, 1236, 189−201. (32) Hori, K.; Matsubara, A.; Uchikata, T.; Tsumura, K.; Fukusaki, E.; Bamba, T. High-throughput and sensitive analysis of 3monochloropropane-1, 2-diol fatty acid esters in edible oils by supercritical fluid chromatography/tandem mass spectrometry. J. Chromatogr. A 2012, 1250, 99−104. (33) Yamazaki, K.; Ogiso, M.; Isagawa, S.; Urushiyama, T.; Ukena, T.; Kibune, N. A new, direct analytical method using LC-MS/MS for fatty acid esters of 3-chloro-1, 2-propanediol (3-MCPD esters) in edible oils. Food Addit. Contam., Part A 2013, 30, 52−68. (34) MacMahon, S.; Mazzola, E.; Begley, T. H.; Diachenko, G. W. Analysis of Processing Contaminants in Edible Oils. Part 1. Liquid Chromatography−Tandem Mass Spectrometry Method for the Direct Detection of 3-Monochloropropanediol Monoesters and Glycidyl Esters. J. Agric. Food Chem. 2013, 61, 4737−4747. (35) MacMahon, S.; Begley, T. H.; Diachenko, G. W. Analysis of Processing Contaminants in Edible Oils. Part 2. Liquid Chromatography−Tandem Mass Spectrometry Method for the Direct Detection

of 3-Monochloropropanediol and 2-Monochloropropanediol Diesters. J. Agric. Food Chem. 2013, 61, 4748−4757. (36) MacMahon, S.; Ridge, C. D.; Begley, T. H. Liquid Chromatography−Tandem Mass Spectrometry (LC-MS/MS) Method for the Direct Detection of 2-Monochloropropanediol (2-MCPD) Esters in Edible Oils. J. Agric. Food Chem. 2014, 62, 11647−11656. (37) Küsters, M.; Bimber, U.; Ossenbrüggen, A.; Reeser, S.; Gallitzendörfer, R.; Gerhartz, M. Rapid and Simple Micromethod for the Simultaneous Determination of 3-MCPD and 3-MCPD Esters in Different Foodstuffs. J. Agric. Food Chem. 2010, 58, 6570−6577. (38) Wöhrlin, F.; Fry, H.; Lahrssen-Wiederholt, M.; Preiß-Weigert, A. Occurrence of fatty acid esters of 3-MCPD, 2-MCPD and glycidol in infant formula. Food Addit. Contam., Part A 2015, 32, 1810−1822. (39) Crews, C.; Chiodini, A.; Granvogl, M.; Hamlet, C.; Hrnčiřík, K.; Kuhlmann, J.; Lampen, A.; Scholz, G.; Weisshaar, R.; Wenzl, T.; Jasti, P. R.; Seefelder, W. Analytical approaches for MCPD esters and glycidyl esters in food and biological samples: a review and future perspectives. Food Addit. Contam., Part A 2013, 30, 11−45. (40) Ermacora, A.; Hrnčiřík, K. Development of an analytical method for the simultaneous analysis of MCPD esters and glycidyl esters in oilbased foodstuffs. Food Addit. Contam., Part A 2014, 31, 985−994. (41) Stamatov, S. D.; Stawinski, J. Regioselective opening of an oxirane system with trifluoroacetic anhydride. A general method for the synthesis of 2-monoacyl- and 1,3-symmetrical triacylglycerols. Tetrahedron 2005, 61, 3659−3669. (42) Ullah, S. M. R.; Murphy, B.; Dorich, B.; Richter, B.; Srinivasan, K. Fat Extraction from Acid- and Base-Hydrolyzed Food Samples Using Accelerated Solvent Extraction. J. Agric. Food Chem. 2011, 59, 2169−2174. (43) AOAC. Official Method 954.02. In Official Methods of Analysis of AOAC International, 18th ed.; AOAC International: Gaithersburg, MD, 2005. (44) AOAC. Official Method 920.39. In Official Methods of Analysis of AOAC International, 18th ed.; AOAC International: Gaithersburg, MD, 2005. (45) Karasek, L.; Wenzl, T.; Ulberth, F. Determination of 3-MCPD esters in edible oil − methods of analysis and comparability of results. Eur. J. Lipid Sci. Technol. 2011, 113, 1433−1442.

J

DOI: 10.1021/acs.jafc.6b04361 J. Agric. Food Chem. XXXX, XXX, XXX−XXX