Anal. Chem. 2004, 76, 4508-4514
Semiautomated High-Throughput Extraction and Cleanup Method for the Measurement of Polybrominated Diphenyl Ethers and Polybrominated and Polychlorinated Biphenyls in Breast Milk Andreas Sjo 1 din,* Ernest E. McGahee, III, Jean-Franc¸ ois Focant, Richard S. Jones, Chester R. Lapeza, Yalin Zhang, and Donald G. Patterson, Jr.
National Center for Environmental Health (NCEH), Division of Laboratory Sciences (DLS), Organic Analytical Toxicology Branch (OAT), Centers for Disease Control and Prevention (CDC), 4770 Buford Hwy NE Mail Stop F-17, Atlanta, Georgia 30341
A semiautomated extraction and cleanup method has been developed to measure eight polybrominated diphenyl ethers (PBDEs), 2,2′,4,4′,5,5′-hexabromobiphenyl (BB153), and 2,2′,4,4′,5,5′-hexachlorobiphenyl (CB-153). The method employs solid-phase dispersion on diatomaceous earth in a solid-phase extraction cartridge followed by automated addition of internal standards (13C-labeled). Extraction is then performed using an automated modular solid-phase extraction system. The extraction procedure includes drying the sample on diatomaceous earth by pressurized nitrogen and eluting target analytes and lipids with dichloromethane. Lipid content is determined gravimetrically. Lipid determinations performed using this method are compared with other standard methods and with a certified reference material. A relative standard deviation of 7.9% was obtained for 130 determinations of the lipid content in a breast milk quality control sample. Final analytical determination of target analytes was performed by gas chromatography-isotope dilution highresolution mass spectrometry. Relative standard deviations for the measurements of target analytes for which a labeled internal standard was available were below 10% for analytes at concentrations above 1 ng/g of lipid. Mean recoveries of the 13C-labeled internal standards ranged from 60 to 89% for the eight PBDE congeners; 74 and 113% were recovered for BB-153 and CB-153, respectively. Flame retardants (FRs) are incorporated into potentially flammable materials such as plastics, rubber, and textiles1-3 to slow the initial phase of a developing fire. Thus, FRs has an * Corresponding author: (e-mail)
[email protected]. (1) WHO. Environmental Health Criteria 162. Brominated diphenyl ethers; International Program on Chemical Safety, WHO: Geneva, Switzerland, 1994. (2) WHO. Environmental Health Criteria 152. Polybrominated Biphenyls; International Program on Chemical Safety, WHO: Geneva, Switzerland, 1994. (3) WHO. Environmental Health Criteria 192. Flame retardants: A general introduction; International Program on Chemical Safety, WHO: Geneva, Switzerland, 1997.
4508 Analytical Chemistry, Vol. 76, No. 15, August 1, 2004
important function in modern society, preventing the development of unintentional fires, limiting the consequences of fires, and hence saving lives. Unfortunately, some FRs such as polybrominated diphenyl ethers (PBDEs) and polybrominated biphenyls (PBBs) are persistent and biomagnify or bioaccumulate in the food chain. PBBs are no longer commercially produced.4 PBDEs are manufactured in three bromination degrees, i.e., technical pentaBDE (triBDEs to hexaBDEs),1,5 technical octaBDE (hexaBDE to nonaBDE),1 and decaBDE containing mainly the perbrominated BDE. All commercial PBDE mixtures are being phased out in the European Union (EU),6 and the pentaBDE and octaBDE mixtures are being voluntarily withdrawn from the U.S. market by the end of 2004. However, the most recent production figures show a substantial consumption of the three PBDE preparations within North America in 2001.7 Worldwide consumption of pentaBDE, octaBDE, and decaBDE in 2001 was 7500, 3790, and 56 100 metric tons, respectively.7 These figures correspond to 95%, 40%, and 44% of the North American demand for the pentaBDE, octaBDE, and decaBDE mixtures, respectively. The concentration of PBDEs (sum of tetraBDE through hexaBDE) in human blood collected in North America is among the highest reported in the world. The average levels of 2,2′,4,4′-tetraBDE reportedly range from 33 to 41 ng/g of lipid,8-10 compared with 2.3 ng/g of lipid in Sweden.11 The exact exposure routes within the United States remain unknown, although it is reasonable to believe that the exposure routes are similar to that of other persistent organohalogen compounds; i.e., food is the dominating route of exposure. (4) de Boer, J.; de Boer, K.; Boon, J. P. In New Types of Persistent Halogenated Compounds; Paasivirta, J., Ed.; Springer-Verlag: Berlin, 2000; Chapter 4. (5) Sjo ¨din, A.; Jakobsson, E.; Kierkegaard, A.; Marsh, G.; Sellstro ¨m, U. J Chromatogr. 1998, 822, 83-89. (6) Cox, P.; Efthymiou, P. Official J. Eur. Union 2003, OJ L 42, 45-46. (7) Bromine Science and Environmental Forum, BSEF. Available: http:// www.bsef-site.com/docs/BFR_vols_2001.doc. (accessed 29 December 2003). (8) She, J.; Petreas, M.; Winkler, J.; Visita, P.; McKinney, M.; Kopec, D. Chemosphere 2002, 46, 697-707. (9) Sjo ¨din, A.; Patterson Jr, D. G.; Bergman, A° . Environ. Int. 2003, 29, 829839. (10) Schecter, A.; Pavuk, M.; Pa¨pke, O.; Ryan, J. J.; Birnbaum, L.; Rosen, R. Environ. Health Perspect. 2003, 111, 1723-1729. (11) Meironyte´, D.; Nore´n, K.; Bergman, A° . J. Toxicol. Environ. Health 1999, 58 Part A, 329-341. 10.1021/ac0495384 Not subject to U.S. Copyright. Publ. 2004 Am. Chem. Soc.
Published on Web 07/03/2004
The aim of the current project was to develop an analytical method for rapid extraction and cleanup of human milk samples to be used in biomonitoring studies designed to assess exposure to the nursing child. We also intended to minimize the sample size and consequently the solvent amounts required per sample. Existing methodologies are based primarily on liquid/liquid extraction approaches using solvents such as hexane and diethyl ether with potassium oxalate as a chelator.10 Our approach was solid-phase dispersion on diatomaceous earth, drying of the sample using pressurized nitrogen followed by elution of the lipid fraction, and a gravimetric lipid determination. This approach is well suited for automation to increase sample throughput. EXPERIMENTAL SECTION Safety. Researchers working with human milk specimens and other potentially pathogen-containing samples must limit their exposure by proper use of personal protective equipment, such as lab coat, protective glasses, and laboratory gloves. Any potential spills of human milk are decontaminated with 10% bleach or 70% ethanol solution, allowing a contact time for decontamination of 15 min. Implementation of a pasteurization step before handling milk samples further decreased the potential risk to the researcher. The researcher also must be aware of the presence of 13C -1,2,3,4-tetrachlorodibenzo-p-dioxin (13C -1,2,3,4-TCDD) in the 6 6 recovery standard used and the potential health effects of exposures to TCDD.12 Certified Reference Standards. Two internal standard spiking solutions obtained from Cambridge Isotope Laboratories (Andover, MA) were used for brominated flame retardants (BFRs) and PCBs, respectively. The spiking standard for BFRs contained the following 13C12-labeled congeners at a concentration of 7.5 pg/ µL in methanol: 2,2,4′-tribromodiphenyl ether (BDE-28); 2,2′,4,4′tetraBDE (BDE-47); 2,2′,4,4′,5-pentaBDE (BDE-99); 2,2′,4,4′,6pentaBDE (BDE-100); 2,2′,4,4′,5,5′-hexaBDE (BDE-153); 2,2′,4,4′,5,6′hexaBDE (BDE-154); 2,2′,3,4,4′,5′,6-heptaBDE (BDE-183); decaBDE (BDE-209); and 2,2′,4,4′,5,5′-hexabromobiphenyl (BB-153). The spiking standard for PCBs contained 2,2′,4,4′,5,5′-hexachlorobiphenyl (CB-153) at a concentration of 7.5 pg/µL in methanol. The recovery spiking standard was also obtained from Cambridge Isotope Laboratories and contained the following labeled compounds: 13C6-1,2,3,4-TCDD, 2.5 pg/µL; 3,3′,4,4′-tetraBDE (13C12BDE-77, 7.5 pg/µL); 2,2′,3,4,4′,6-hexaBDE (13C12-BDE-139, 7.5 pg/ µL); and 2,2′,3,3′,4,5,5′,6,6′-nonaCB (13C12-CB-208, 10 pg/µL). The solvent for the recovery standard was n-hexane containing 10% and 2 vol % nonane and dodecane, respectively. A 10-point calibration curve spanning the concentration range 0.2-2000 pg/µL containing the 13C-labeled BFRs and CB-153 at a concentration of 75 pg/µL was used for the gas chromatography/ isotope dilution high-resolution mass spectrometry (GC/IDHRMS) analyses. The calibration curve contained the following native analytes: 2,2′,4-triBDE (BDE-17); BDE-28; BDE-47; 2,3′,4,4′tetraBDE (BDE-66); 2,2′,3,4,4′-pentaBDE (BDE-85); BDE-99; BDE100; BDE-153; BDE-154; BDE-183; 2,2′,3,4,4′,5,5′,6-octaBDE (BDE203); BDE-209; BB-153; and CB-153. Chemicals. Reagents and solvents used in the current method were of the highest possible grade available, intended for pesticide residue analysis and were used only after verification by GC/ (12) Myers, G. L.; Patterson Jr, D. G. Professional Saf. 1987, 32, 30-37.
IDHRMS analysis monitoring for the BFRs listed above. Dichloromethane (DCM), n-hexane, and methanol were of pesticide grade and purchased from TEDIA (Fairfield, OH); water of HPLC grade also was purchased from TEDIA. Chloroform was obtained from EM Scientific, an affiliate of Merck (Darmstadt, Germany). Diatomaceous earth (Hydromatrix) was obtained in bulk from Varian (Walnut Creek, CA) and was cleaned by accelerated solvent extraction (ASE) in an ASE200 instrument (Dionex; Sunnyvale, CA) with DCM at 100°C, using a static time of 5 min, a 30% purge volume, and two extraction cycles. The diatomaceous earth was then stored in a Thelco laboratory oven (Precision; Winchester, VA) kept at 250 °C. SPE cartridges (3 mL) containing diatomaceous earth (0.8 g) were prepared inhouse at time of use. Custom-made solid-phase extraction (SPE) cartridges containing activated silica (0.1 g; top layer) and silica/sulfuric acid 2:1 by weight (1 g; bottom layer) were obtained from Applied Separations (Allentown, PA). Further information about the development of these custom-made cartridges is referenced.13 Cleaning of Glassware and Other Consumables. All glassware was cleaned in a dishwasher steam scrubber (Labconco; Kansas City, MO) and heated in a Thelco laboratory oven (Precision; Winchester, VA) at 250 °C overnight, before use, to remove any contaminants. Test tubes (16 × 100 mm with treads for screw caps) were obtained from Fisher Scientific (Suwanee, GA), and tubes (16 × 100 mm with no treads) were obtained from Corning (New York, NY). Pasteur pipets were not washed in the dishwasher but burned directly in the same oven overnight, to remove any contaminants from the glass. Teflon-lined screw caps and Teflon-lined silicone septums (Wheaton; Millville, NJ) were sonicated in methanol and air-dried before use. After cleaning, all glassware and caps were stored in a closed cabinet or drawer in Ziploc bags or in beakers covered with aluminum foil. Breast Milk Samples. Pooled human milk used to develop the analytical method was obtained from the Mothers’ Milk Bank (Denver, CO). Upon arrival, the milk was transferred to a -70 °C freezer for storage. To minimize freezing and refreezing, the milk was aliquoted into 10-mL glass flasks equipped with Teflonlined screw caps. The milk pool used was a composite pool of two women collected in 2002. Bovine milk diluted 1:10 with deionized water obtained in a local supermarket was used as method blank samples. Two pools of human milk pools (10 donors per pool) from California and North Carolina also were available to us. The samples were collected by lactation centers, located in each state during December 2002 through January 2003. All women were healthy and had healthy infants, were breast-feeding only one infant at the time, and had not lived outside of their location for more than 6 months in the last 5 years. Each woman collected 50 mL of milk with a breast pump after making sure their breasts and hands were clean and free of ointments. The milk at each lactation center was pooled to form two specimens, representing California and North Carolina. Milk specimens were stored in precleaned glass containers with Teflon-lined inner lids and kept frozen (-25 °C) during collection and transport to the laboratory. Potassium dichromate (0.1%, w/w) was added to the milk (13) Sjo ¨din, A.; Jones, R. S.; Lapeza, C. R.; Focant, J.-F.; McGahee, E.; Dublin, G.; Patterson Jr, D. G. Organohalogen Compd. 2003, 60, 412-415.
Analytical Chemistry, Vol. 76, No. 15, August 1, 2004
4509
specimens to prevent biological growth. The project protocol was determined not to be human subjects research because the milk specimens were pooled and no personal identifying information was collected. Automated Sample Pretreatment. Addition of internal standards before extraction was automated using a 215 Liquid Handler (Gilson; Middleton, WI) fitted with a 402 syringe pump (Gilson). The syringe pump was equipped with a 250-µL syringe as well as a 10-mL syringe (Gilson). The syringe pump was connected to a septum-piercing probe via 10.5 mL of coiled-transfer tubing. A methanol reservoir was connected to the syringe pump to purge the transfer tubing and to rinse the septum-piercing probe (Gilson) used for drawing and dispensing the standard solutions. The above instrumentation was controlled and operated under the UniPoint v5.5 (Gilson) software. Automated SPE and Cleanup. SPE and cleanup were automated using the Rapid Trace SPE Workstation (Zymark; Hopkinton, MA). The Rapid Trace is a modular SPE system. The above instrumentation was controlled and operated under the Rapid Trace workstation software v1.20 (Zymark) software. Sample Evaporation. A Rapid Vap (Labconco; Kansas City, MO) was used for sample evaporation employing vortex action, vacuum, and elevated temperature to aid the evaporation process. Sample and Lipid Weight Determination. The amount of lipids in the extracts was determined gravimetrically with an analytical balance AX105 Delta Range (Mettler Toledo; Columbus, OH) with an accuracy of (0.01 mg. The original milk samples were weighed using a conventional laboratory balance Adventurer (accuracy (0.001 g) (Ohaus; Pine Brook, NJ). Isotope Dilution High-Resolution Mass Spectrometry. IDHRMS analysis was performed on a MAT95XP (ThermoFinnigan MAT; Bremen, Germany) instrument. The chromatographic separations were carried out on a 6890N gas chromatograph (GC) (Agilent; Atlanta, GA) fitted with a DB5HT (15 m, 0.25-mm i.d., and 0.10-µm film thickness; Agilent) capillary column. Splitless injection was carried out with an injector temperature of 280 °C; the oven was programmed from 140 (1 min) to 320 °C (0 min) with a ramp rate of 10 °C/min. The source temperature was 280 °C and operated in the electron impact mode using a filament bias of 40 eV. The selective ion monitoring descriptor used has previously been published.13 Pasteurization of Milk before Analysis. The proposed extraction method is a nondestructive method; hence, all milk samples were pasteurized to eliminate the risk for exposure to viruses (such as HIV) and bacteria. Holder pasteurization (62.5 °C for 30 min) is used for pasteurization of breast milk collected by donor banks for human consumption.14 We adopted the same pasteurization for the proposed method but used a laboratory oven instead of a water bath to heat the samples. The pasteurization was performed as follows. Vials were placed in the laboratory oven heated to 65 °C. The samples required 60 min to reach pasteurization temperature and then were left in the oven for another 60 min to ensure complete pasteurization. The effect on lipid content of the milk was further determined at several time points, i.e., no pasteurization and 0.5, 2, and 12 h. No substantial effect on lipid content was observed for samples pasteurized for 0.5 and 2 h, cf. Figure 1. (14) Tully, D. B.; Jones, F.; Tully, M. R. J. Hum. Lact. 2001, 17, 152-155.
4510 Analytical Chemistry, Vol. 76, No. 15, August 1, 2004
Figure 1. Influence of pasteurization on the determination of lipids in human milk by the proposed method compared to unpasteurized milk. Pasteurization times were 0.5, 2, and 12 h at 65 °C. Pasteurization according to Holder14 was performed at 62.5 °C for 0.5 h. Significant differences (t-test) are indicated (*; p < 0.05).
Quality Assurance and Quality Control (QA/QC). In this method, a set of samples was defined as eight unknown samples, prepared, and analyzed together with one analytical blank and one QA/QC sample. Two sets were prepared and analyzed at the same time as one batch. For QA/QC purposes, measurement of a target analyte in a set of samples was considered valid only after the QA/QC sample had fulfilled the following criteria: (1) the measurement of the target analyte in the QA/QC sample must not fall outside the interval defined as (3 standard deviations of the established mean of the QA/QC samples and (2) 10 or more consecutive measurements of the QA/QC sample must not fall above or below the established mean of the QA/QC samples after one QA/QC sample has failed criterion 1. Furthermore, every measurement of a set of samples must fulfill the following criteria to be considered a valid measurement: (1) the ratio of the two ions monitored for every analyte and 13C-labeled internal standard must not deviate more then 20% from the theoretical value; (2) the ratio of the retention time of the analyte over its corresponding 13C-labeled internal standard must be within the range of 0.99-1.01. For analytes that did not have an identical 13C -labeled IS, the ratio to the IS used could not deviate more than 1% from the average of the same ratio of the calibration standards analyzed in the same analytical run; and (3) the measured recovery of the IS must be within the range 25-150%. Sample Pretreatment. SPE cartridges containing diatomaceous earth were prepared in-house immediately before use. The diatomaceous earth was previously cleaned by ASE (dichloromethane; 100 °C) and stored in the laboratory oven at 250 °C. The milk to be extracted was added directly to the SPE cartridge before a polyethylene frit was added on top of the diatomaceous earth. The amount of milk added (∼1 g) to the cartridge was determined by laboratory balance. The balance was tarred with the SPE cartridge containing the diatomaceous earth, and the amount added was then recorded after the addition. During the addition of the milk and the addition of internal standard, the SPE cartridge was placed in a disposable 16 × 100 mm test tube. The cartridge is supported by the wings of the cartridge and does not touch the bottom of the test tube. After addition of the milk to the cartridges, the samples are spiked with internal standards for the BFRs and CB-153 using an automated liquid handler. The test tubes holding the SPE cartridges are placed in a rack for 16 × 100 mm test tubes in the instrument. GC vials also were loaded into a separate custom-made rack on the liquid handler. Sample
processing was initiated through the computer software. We first purged the probe and transfer tubing over a rinse station by drawing methanol (10 mL) from the reservoir and rinsing both the outside and the inside of the probe. This also ensured that no air was present in the transfer tubing. We then added recovery standard (100 µL) to the empty GC vials using the command “dilute and add”, in which first an air gap (40 µL) was drawn followed by the recovery standard. The air gap separated the standard solution from the methanol in the transfer tubing. The standard was then dispensed into the GC vial, and 50 µL of methanol was dispensed from the reservoir, ensuring that all of the standard had entered the vial. After completing this step, we capped the GC vials and placed them in a refrigerator until the end of the cleanup procedure. The internal standards for BFRs and PCBs were added to the milk samples in the SPE cartridges using the same spiking command. The liquid handler processed unattended 20 samples in ∼20 min. Extraction. The samples were extracted unattended using the automated SPE workstation located inside a fume hood. The method included drying the milk samples by pressurized nitrogen for 120 min at a pressure of 10 psi and elution of the cartridge with dichloromethane (12 mL). During the drying step, the SPE cartridge was located over an empty collection tube for visual verification that no milk was forced from the SPE cartridge during the drying. At the pressure used, this did not occur. After the drying step, the samples were eluted with dichloromethane into another 16 × 100 mm test tube, which had threads for screw caps. This tube was weighed earlier on an analytical balance for later gravimetrical determination of coextracted lipids. Five samples were loaded onto each module; hence, a set of samples was divided over two modules, with the blank and QA/QC sample processed on different modules. During normal processing, two sets are processed simultaneously using four modules. Automated processing time using four modules for a batch of 20 samples took 11.5 h (35 min/sample). Determination of Lipid Content. After extraction, the samples were evaporated to dryness using the vacuum evaporator at 550 mbar, 40 °C, and a vortex action of 40%. After complete evaporation of the solvent, the pressure setting was reduced to 230 mBar for an additional 30 min and the amount of coextracted lipids was determined gravimetrically on an analytical balance. The evaporation was continued for an additional 20 min, and the weights were recorded again for verification of complete evaporation of the solvent. Removal of Coextracted Lipids. The resulting extracts were then eluted through a two-layered SPE cartridge (3 mL) packed with silica (0.1 g; activated 250 °C; top layer) and silica/sulfuric acid (1 g; 2:1 by weight; bottom layer); the two layers were separated by a polypropylene frit. The samples and method blanks were processed unattended for cleanup using the automated SPE instrument. The cartridges used for sample cleanup were first conditioned with hexane (10 mL). The syringe pump on the instrument then delivered hexane (1 mL) to the tube containing the sample extract. Drawing the hexane back into the syringe pump and then back into the test tube twice ensured complete reconstitution of the sample. The syringe pump then delivered the sample to the cleanup cartridge. This procedure was repeated
twice with additional hexane (2 × 1 mL) to ensure a quantitative transfer. We eluted the sample by passing an additional 8 mL of hexane through the cartridge. A total volume of 11 mL of hexane was collected during this procedure into 16 × 100 mm test tubes. A processing time per sample of 30 min was obtained; simultaneous processing of five samples per module gave a processing time of 2.5 h for a set of 20 samples. After evaporation, the extracts were manually transferred to the GC vials that were spiked with recovery standard (cf. sample pretreatment) and were evaporated to a volume of 10 µL using the vacuum evaporator at 40 °C and 230 mbar with a vortex action setting of 40%. The nonane and dodecane present in the recovery standard acted as a keeper for volatile components. Reference Methods of Lipid Weight Determination. To validate the lipid weight determinations of the developed methodology, we employed two reference methods. The extraction method by Bligh and Dyer15 was used with modifications (see below), and the method by Hovander et al.16 was used in accordance with the original reference, although this method was not intended for milk. We performed lipid extraction according to Bligh and Dyer,15 as follows. Milk (1 g) was weighed into 16 × 100 mm test tubes. Chloroform (1 mL) and methanol (2 mL) were added to the samples, followed by rigorous mixing on a vortex mixer. Additional chloroform (1 mL) and water (1 mL) were added with rigorous mixing between the additions. The sample was then centrifuged to aid the phase separations. The organic layer was transferred to a second test tube. This procedure was repeated two more times with additional chloroform (1 mL) to ensure a quantitative transfer of the organic layer. Because of observed protein residues in the final extract, the sample was evaporated and reconstituted in hexane and partitioned against a 0.1 M hydrochloric acid solution. The lipid weights were then gravimetrically determined. RESULTS The amount of lipids recovered for the certified bovine milk sample and locally obtained bovine milk differed significantly from results of the proposed methodology and reference methods15,16 (Table 1). Lipid concentration determined in the human milk using the proposed methodology did not differ significantly from determinations using the two reference methods.15,16 No significant effect of pasteurization (t-test) of the human milk was observed at a pasteurization time of 30 min, although a decrease in lipid content was observed after heating the samples to pasteurization temperature for 2 h or more (Figure 1). Overall mean recovery of the labeled internal standards BDE28, BDE-47, BDE-99, BDE-100, BDE-153, BDE-154, BDE-183, BDE-209, CB-153, and BB-153 were 83% (SD13), 89% (SD15), 75% (SD10), 71% (SD10), 80% (SD17), 73% (SD11), 82% (SD17), 60% (SD24), 113% (SD40), and 74% (SD12), respectively. The relative standard deviation (RSD) for the quantitative measurements of all target analytes in the milk when an identical 13C label was available and the concentrations in the pool were above 1 ng/g lipid ranged from 3.6% to 8.7% depending on the analyte (Table 2). QA/QC charts for the determined lipid concentration and (15) Bligh, E. G.; Dyer, W. J. Can. J. Biochem. Physiol. 1959, 37, 911-917. (16) Hovander, L.; Athanasiadou, M.; Asplund, L.; Jensen, S.; Klasson-Wehler, E. J. Anal. Toxicol. 2000, 24, 696-703.
Analytical Chemistry, Vol. 76, No. 15, August 1, 2004
4511
Table 1. Lipid Concentration As Determined in Two Samples of Bovine Milk and One Pool of Human Milk by the Proposed Extraction Method and Two Reference Methodsa proposed method solid-phase dispersion milk sample milkb
human breast certified bovine poolc locally obtained bovine milkd
Hovander et al.16 method
Bligh and Dyer method15
average (%)
RSD
average (%)
RSD
p-value
average (%)
RSD
p-value
4.3 (n ) 131) 3.6 (n ) 10) 1.6 (n ) 5)
3.6 7.6 7.8
4.1 (n ) 10) 4.0 (n ) 10) 1.9 (n ) 10)
7.3 4.5 2.4
0.3
