Anal. Chem. 1995, 67,4155-4163
An Integrated Analytical Method for Determination of Polychlorinated Aryl Methyl Sulfone Metabolites and Polychlorinated Hydrocarbon Contaminants in Biological Matrices Robert J. Letcher and Ross J. Norstrom* Centre for Analytical and Environmental Chemistry, Department of Chemistty, Carleton University, Ottawa, Ontario, KIS 566 Canada, and National Wildlife Research Centre, Canadian Wildlife Service, Hull, Quebec, KIA OH3 Canada
h e Bergman Department of Environmental Chemistty, Wallenberg Laboratory, Stockholm University, S-10691 Stockholm, Sweden
A simple column chromatography method was developed for separation and cleanup in the determination of chlorinated hydrocarbon contaminants and their methyl sulfone (MeSOz-) metabolites in biological tissues. The method was validated for determination of 11polychlorinated biphenyls (PCBs), 15tetra- to heptachlor0 3-and 4-MeS02-PCBs,3-MeS02-DDE,and tris(4-chlor0phenyl)methanol spiked to herring gull egg,smelt, and polar bear liver and adipose tissue using gas chromatographywith electron-capture detection (GC-ECD). The overall mean recovery relative to the internal standardwas 103%f8%, independent of analyte, substrate type, and lipid extract weights up to -0.7 g. Precision of replicate analyses of individualcongenerswas good. There were no sigdfkant residual biogenic or xenobiotic interferences in the aryl methyl sulfone fraction of any substrate. Sensitivity and linearity of molar response of MeS02-PCBs and MeS02DDE was tested for ECD and electron-capture negative ion mass spectrometry monitoring the total ion current (TIC) and the molecular ion (SIM). The mean practical quantitationlimit among MeS02-PCBsand 3-MeS02-DDE was lowest for SIM (2.1 f 0.9 pg) and similar for ECD and TIC (24.2 k 4.6 and 44.4 f 17.1 pg, respectively). Response factors were linear above the practical quantitation limit to at least the nanogram level for all three techniques. In spite of superior sensitivity, there was more inherent variability in the response factors for SIM (-l8%-56%CV) than for ECD (-7%-12% CV) or TIC (-ll%-18% CV);therefore, ECD orTIC is recommended for quantitative analysis. Chlorinated hydrocarbon contaminants (CHCs) such as polychlorinated biphenyls (PCBs) and 1,ldichloro-2,2-bis(4-chloropheny1)ethene (4,4'-DDE), a metabolite of the pesticide DDT, are well-known global environmental pollutants that bioaccumulate in biota.' In contrast, methylsulfonyl- (MeS02-) containing
'Address correspondence to this author at Environment Canada, Canadian Wildlife Service, National Wildlie Research Centre, 100 Gamelin Blvd., Bldg. #9, Hull, QuCbec, KlA OH3 Canada. Telephone: (819) 997-1411. Fax: (819) 953-6612. E-mail:
[email protected]. 0003-2700/95/0367-4155$9.00/0 0 1995 American Chemical Society
metabolites of PCBs and DDE have received scant attention as environmental contaminants since their first discovery in Baltic seals in 1976.2 Aryl methyl sulfone metabolites have escaped detection in standard analysis procedures, as they are more polar relative to other common CHCs and quantitation standards have not been readily available. Knowledge of MeSOTPCB toxicology is on the rise but remains poor relative to the parent c0mpounds.3~~ MeSOTDDE is a potent adrenocortical toxicant in some specie^.^^^ MeSOzPCBs are slightly less hydrophobic than the parent PCBs; however, MeSOTPCBs possess lipophilic properties similar to those of other bioaccumulating CHCS.~MeSOTPCBs fulfillthe requirements for bioaccumulation since they have octanol/water partition coefficients (log Kow)greater than 3. Furthermore, bioaccumulated MeSOzPCBs and -DDE appear to resist further metabolic degradation. At least 22 MeSOzPCB and two MeSOT DDE compounds at total levels ranging from 2% to 20% of total PCB and DDE levels have been identified in mammalian tissue, including Aryl methyl sulfone compounds possess Lewis basicity, acid/ base stability, and high polarity, which are useful properties for separation from coextracting lipids and other CHCs in biological tissues. Existing analytical methodologiesfor aryl methyl sulfone (1) Tanabe, S.; Iwata, H.: Tatsukawa, R Sci. Total Enuiron. 1994,154,163177. (2) Jensen, S.; Jansson, B. Ambio 1976,5, 257-260. (3) Kato, Y.; Haraguchi, IC; Kawashima, M.: Yamada, S.; Masuda, Y.; Kimura, R Chem.Bio1. Interact. 1994,95, 257-263. (4) Kiyohara, C.; Hirohata, T. Toxicol.in Vitro 1994,8, 1185-1189. (5) Jonsson, C.-J.; Lund, B.-0.; Brunstrom, B.; Brandt, I. Enuiron. Toxicol.Chem. 1994,13,1303-1310. (6) Lund, B:O. Enuiron. Toxicol. Chem. 1994,13,911-917. (7) Bergman, A; Haraguchi, IC Cambridge Isotope Laboratories Supplemenf; Cambridge Isotope Laboratories: Cambridge, MA, 1994; Vol. 1, pp 1-8. (8) Letcher, R J.; Norstrom, R J.: Bergman, A Sci. Total Enuiron. 1995,160, 409-420. (9) Bergman, A; Norstrom, R J.; Haraguchi, IC;Kuroki, H.; Beland, P. Enuiron. Toxicol. Chem. 1994,13,121-128. (10) Haraguchi, IC;Kuroki, H.; Masuda, Y.; Shigematsu, N. Food Chem. Toxicol. 1987,22, 283-288. (11) Haraguchi, IC; Kuroki, H.; Masuda, Y.J. Chromatogr. 1986,361,239252. (12) Bergman, bL; Athanasiadou, M.; Bergek, S.; Haraguchi, IC; Jensen, S.; Klasson-Wehler,E. Ambio 1992,21,570-576. (13) Haraguchi, IC; Athanasiadou, M.; Bergman, A; Hovander, L.; Jensen, S. Ambio 1992,21, 546-549.
Analytical Chemistry, Vol. 67, No. 22, November 15, 1995 4155
cleanup and/or CHC separation have exploited these physicoTable 1. MeSOTPCB, MeSOTDDE, and Internal chemical characteristics by incorporation of liquid-liquid partiStandard Chemical Abbreviatlons tioning strategie~.2%~-~* Dimethyl sulfoxide, concentrated sulfuric GC peak no. substituted biphenyl compound acid, and acetonitrile partitioning have been common approaches. in Figure 2u3b structural formula abbreviationC Lipiddestructive tissue extraction strategies, such as saponification 1* 3-CH3SOp2,2’,4‘,5-C& 5MeSOpCB49 with alcoholic potassium hydroxide at elevated temperatures,have 2 4CH3SOp2,2’,5,5’-C& 4-MeSOeCB52 3* 4-CH3SOp2,2’,4‘,5-C14 4MeSOpCB49 also been used.I4 To simplify the procedures and enhance 4 3-CH3SOp2,4‘,5,6-C& 5MeSOpCB64 reliability of determination, we developed a simple column 5* 4-CH3SOp2,4‘,5,6-C& 4MeSOpCB64 chromatography approach amenable to routine CHC and aryl 6 3-CH3SOp2,3’,4‘,5-C& SMeSOpCB70 7* 3-CH3SOp2,2’,4‘,5,5’-Cls 5MeSOpCB101 methyl sulfone isolation from biological tissue samples. The 8 4-CH3SOr2,3’,4‘,5-C4 4MeSOpCB70 method takes advantage of aryl methyl sulfone polarity, which is 9* 4CH3SOp2,2’,4‘,5,5’-Cls 4-MeSOrCB101 higher than that of most other biologically persistent CHCs, such 10* 3-CH3SOpDDE 3-MeSOpDDE 11* 3-CH3SOp2,2’,3’,4‘,5(315 3-MeSOpCB87 as PCBs and chlorinated insecticides (e.g., DDTs and chlordanes), 12* 3-CH3SOp2,2’,4’,5,5’,&cl6 5MeSOpCB149 and the smaller molecular size of aryl methyl sulfones and CHCs 13 4-CH3SO22,2’,3’,5,5’,6‘-CI6 4MeSOrCB151 than biogenic coextracting compounds, mainly triglycerides. Gas 14* 4-CH3SOp2,2’,3’,4‘,5-C15 4MeSOpCB87 15* 4CH3Sop2,2’,4‘,5,5’,&Cl6 4-MeSOpCB149 chromatography/electroncapture detection (GC-ECD) was used 16* 3-CH3SOp2,2’,3’,4‘,5,&cl6 3-MeSOpCB132 to determine the percent recovery of (1) standard mixtures of 17* 4-CH3SOp2,2’,3’,4‘,5,&C& 4-MeSOzCB132 tetra- to heptachloro-3- and -4-MeSOrPCBs, and 3-MeSOrDDE, 18* 3-CH3SOp2,2’,3’,4‘,5,5’-C16 SMeSOpCB141 19* 4-CH3SO22,2’,3’,4‘,5,5’-CI6 4-MeSOrCB141 which are biologically ~ignificant,~’~ and an internal standard (IS) 20* 3-CH3SOp2-CH3-2’,3’,4‘,5,5’-C15 IS of 3-MeS0~2-methyl-52’,3’,4‘,5’-pentachlorobiphenyl spiked to 21* 3-CH3SOr2,2’,3’,4‘,5,5’,&C17 3-MeSOpCB174 herring gull egg, whole fish (smelt), polar bear liver, and polar 22* 4-CH3SOp2,2’,3’,4‘,5,5’,&C17 4-MeSOrCB174 3-CH3SOr2,2’,5,5’-C4 5MeSOpCB52 bear adipose tissue homogenates and (2) the same MeSOrPCBs, 4-CH3SOp2,2’,5,5’,6-C15 4MeSOrCB95 3-MeSOrDDE, IS, dichloro- to octachloro-PCBs, and tris(4 3-CH3!30~2,2’,4‘,5,6-C15 SMeSOpCB91 chloropheny1)methanol (TCPMeOH) spiked to contaminant-free 3-CH3SOp2,3’,4‘,5,6Cl~ 5MeSOpCB110 4-CH3SOp2,3’,4‘,5,6-C15 4-MeSOpCB110 lipid extracts of the same samples prepared by gel-permeation chromatography (GPC). TCPMeOH has recently been shown to Standards for all compounds, except 4, 6, and 13, were used for be a ubiquitous global environmental ~ o n t a m i n a n t . ~The ~J~ GC-FID concentration verification. Com ounds with asterisks were used for all other studies (Tables 2-4). Congeners 19, 21, and 22 polarity of TCPMeOH is at the high end of routinely determined were not spiked to hemng gull egg, and congeners 12, 14, and 16 CHCs, such as dieldrin, and separation from MeSOrPCBs and were not spiked to polar bear liver and adipose tissue for percent recovery determinations (Table 2). MeSOpPCB congeners not num-DDE is a good test of the method. bered have not been identiiied in environmental samples to date. The aryl methyl sulfone fractions obtained from tissue homoBased on IUPAC numbering of precursor PCB congener.’* CB stands for chlorobiphenyl. genates were analyzed by GC-ECD as well as by gas chromatography/electron-capture negative ion mass spectromeby (GC/ ECNI-MS) to evaluate residual biogenic and xenobiotic interferences. GC/ECNI-MS has been used for MeSOrPCB chlorievaluated to determine if isomer group RMRFs could be used for nated isomer group and MeSOrDDE characterizati~n*~~J~J~ but MeSOrPCB and -DDE quantitation in environmental samples. not as a quantitative technique. Sensitivity of MeSOrPCBs and -DDE in GC-ECD and GC/ECNI-MS in the selected ion monitorEXPERIMENTAL SECTION ing (SIM) mode is in the low picogram range, as these compounds Chemicals and Standards. MeSOrPCBs, SMeSOpDDE, are electrophilic and readily capture thermally energized elecand the internal standard (IS), 3-MeS022-methyl-52’,3’,4‘,~t r o n ~ . In ~ this ~ , ~study, ~ the response characteristics among ECD, pentachlorobiphenyl, were synthesized as described previou~ly.~~ SIM, and TIC were compared to determine which techniques were The MeSOrPCB and -DDE metabolite chemical names were the best for precise and accurate MeSOpPCB and -DDE quantiabbreviated and simplified (Table 1) on the basis of the IUPACtation. Technique sensitivity was evaluated by comparing the derived numbering system of the parent PCBs.l8 The PCB instrumental detection limits (IDLs) of 15 MeSOzPCBs and standard mixtures, CLB-1-Dand CLB-14, were obtained from the 3-MeSOz-DDE to the corresponding method detection limits National Research Council (NRC) of Canada. TCPMeOH was (MDLs) and practical quantitation limits (PQLs) determined from purchased from MTM Research Chemicals (Lancaster Synthesis CHC-free lipid extracts of polar bear liver. The stability of ECD, Inc., Wmdham, NH). All solvents were of analytical grade or SIM, and TIC molar response factors (MRFs) for the same MeSOp better. PCBs and 3-MeSOrDDE were determined over a mass range from Verification of Standard Concentrations. Two standard -1 to 2 ng down to PQL levels. The constancy of the relative mixtures (nominal concentration 60 ng pL-I per congener), one MRFs (RMRFs) to an internal standard among congeners was with 15 MeSOrPCBs (Table l),IS, and 11dichloro- to octachloroPCBs (NRGCLB-1-C) and the other with seven MeSOrPCBs (14) Haraguchi, K.;Kuroki, H.; Masuda, Y.1. Anal. Toxicol. 1 9 8 4 , 8, 177-181. (Table l),SMeSOrDDE, IS, and the NRC-CLB-1-Csolution, were (15) Zook, D. R ; Buser, H.-R; Bergqvist, P.-A;Rappe, C.; Olsson, M. Ambio 1991,21,557-560. analyzed by oncolumn injection GC with 5ame ionization detection (16) Jarman, W.M.; Simon, M.; Norstrom, R. J.; Bums, S. A; Bacon, C. A; (GC-FID) to verify the concentration of the standards. GC-FID Simoneit, B. R T.; Riseborough, R W. Environ. Sci. Technol. 1 9 9 2 , 26, conditions are given below. FID response of halogenated hydro1770-1774. (17) Haraguchi, K; Bergman, A; Jakobsson, E.; Masuda, Y. Freseniw J. Anal. carbons, even those containimg heteroatoms, has been shown to Chem. 1993,347, 441-449. be relatively constant when adjusted for the fraction of the (18) Ballschmiter, K; Mennel, A,; Buyten, J. Fresenius J. Anal. Chem. 1 9 9 3 , molecular weight that is carbon.*g~*OTwo standard mixtures were 346, 396-402. 4156
Analytical Chemistty, Vol. 67, No. 22, November 15, 1995
prepared to ensure complete GC resolution of the MeS02PCBs, SMeSOTDDE, and IS, and the molar carbon response factors (MCRFs) were determined in triplicate. The mean MCRF among the 11 PCB congeners in the primary standard was 5.89 f 1.15 peak area counts pg-1 injected total carbon wt-I of the molecular wt. The mean MCRF of the 22 MeSOrPCB congeners, 3-MeSOr DDE, and IS (Table 1) was 5.57 f 1.98 peak area counts pg-I injected total carbon wt-1 of the molecular wt. MCRFs of 3-MeSOTDDE and five'out of the 22 MeSOrPCB standards fell outside the mean f SD range for PCBs. If these five MeSOz PCBs and 3MeSOTDDE were excluded, the mean MCRF of the remaining 17 MeSOrPCB and IS standards was 5.82 f 1.05, the same as that of the PCBs. Therefore, the nominal concentrations of the six outliers were adjusted by multiplying by the ratio of their MCRFs by the average MCRF of the PCB and other MeSOr PCB standards. Column Chromatographic Materials and Apparatus. Florisil (magnesium silicate, BDH Chemical Inc., Toronto, Canada; pesticide analysis grade, 60-100 mesh) was activated at 600 "C overnight, cooled to 100 "C, deactivated with 1.2%(w/w) doubly distilled, n-hexane-washed water, and stored in a capped (Teflonlined) glass bottle. Basic aluminum oxide (Fisher Scientific Inc., Ottawa, Canada; activity grade I, 60-325 mesh) and silica gel @ioRad Laboratories, Richmond, CA; Bio-si1A, 100-200 mesh) were activated at 300 "C and 180 "C, respectively, for 3 h. The activated materials were stored at 100 "C in an open top glass bottle for no more than 1 month before reactivation. The basic alumina was deactivated with 2.3% (w/w) doubly distilled, n-hexane-washed water 0.5 h prior to use. Potassium hydroxide (KOH>/silica gel (33%KOH (w/w)) was prepared by combining 1M KOH (Fisher Scientific Inc.; prepared with n-hexane-washed, doubly distilled water) with the appropriate amount of activated silica gel 0.5 h prior to use. Sodium sulfate (BDH Chemical Inc.; anhydrous, granular,analyticalgrade) was washed with 5050 &chloromethane/ n-hexane, air-evaporated overnight,activated at 650 "C Overnight, cooled to room temperature, and stored in a Teflon-lined,capped bottle. GPC was performed using an ABC Laboratories automated system. A 2.5 cm diameter column was packed with 60 g of SX-3 Envirobeads (ABC Laboratories, Columbia, MO; 200-400 mesh) after presoaking -1 h in 5050 dichloromethaneln-hexane. Pyrex glass columns 1.0 cm in diameter were used for all liquid chromatography. Tissue Sample Preparation, Standard Spiking, and Fxtraction. Samples archived at -40 "C in the specimen bank at the National Wildlife Research Centre in Hull, PQ, Canada, were used for method validation. Lake Ontario (1989) herring gull (Lam argentatus) egg and Lake Huron (1985) smelt (Osmem mordax) samples were obtained from composites of 118 and 300 individual samples, respectively. Polar bear (Vnus maritimus) liver homogenate and minced adipose samples were from an adult male collected in 1993 from the Resolute Bay area of the Northwest Territories, Canada. Homogenized or minced samples were dehydrated by g r i n h with sodium sulfate in a 1:5 weight ratio. Tissue substrates for method validation were spiked with one of three standard mixtures (Table 1;Figure 1,position A) consisting of 15 or 12 MeSOrPCBs, SMeSOrDDE, and IS (Table 1). Spiked MeSOrPCB and -DDE levels (50 p L of 52-552 pg pL-l) approximately matched native ~
~
(19)Rem, H.;Qingyu, 0.; Weile, Y. Anal. Chem. 1990,62, 2063-2064. (20) Tong, H. Y.;Karasek, F. W. Anal. Chem. 1984,56, 2124-2128.
TISSUE SUBSTRATE
I I
1. homcgrnize 2. rodlum suhte
Extraction 50:50 CH2Cl2ln-hexane
Lipid Extract
@ CHC.fm llpld oxtnct (Fl) I---
l
33% KOH/Silica Gel 5050 CHpClpln-hexane(Fl)
1.2% H 2 0 deact. Fiorisil 5050 CH2C12/n-hexsnr
r
CHC8 (FI)
(GCIEI-MS)
7:93 MIOH/ CHzCl2 (FZ)
2.3% H 2 0 deact. Basic Alumina
5050 CH~Clp/n-hexane
50:50 CH2CI2/n-hexano discard (F')
I
Awl MOUty/ (F2) Sulfones (GCIECNI-MS + GCIECD)
Clgure 1. Schematic diagram of column chromatography methodology. MeS02-PC6, 3-MeSOz-DDE, and IS standards were spiked at point A and MeS02-PC6, 3-MeSOz-DDE, IS, PCB, and TCPMeOH standards at point B (Table 1) for method validation. The part of the schematic in dashed lines was for method validation with contaminantfree lipid extracts only. Extraction volumes are given in the Contaminant Enrichment and Cleanup section.
levels in 0.5 g wet wt of polar bear liver and adipose tissue to minimize error due to subtractionof GGECD peak areas. Native levels in herring gull egg (5.0 g) and smelt (15.0 g) were low ng g-1 wet wt level and nondetectable, respectively, and therefore did not require spiking concentration adjustment. Herring gull egg and smelt homogenates were spiked with the same concentrations of MeSOrPCB, 3-MeSOrDDE, and IS standards used for contaminant-free lipid extracts (see below). Lipid extracts were obtained by column extraction with 80 mL of 50:50 dichloromethaneln-hexane. The dump cycle of gel-permeation chromatography (GPC, see next section) was essentially free of CHCs. However, CHCs such as chlorinated paraffins are similar in molecular size to long chain lipids.21 GPC separation is based largely on size exclusion, and therefore chlorinated paraffins, if present in the sample, coelute with the lipid extract. The lipid extract was considered "contaminant-free" since remaining chlorinated paraffins would be removed by the second GPC loop. The GPC dump cycle was spiked with 50 p L of a CLB-1-D PCB (30295 pg pL-') and TCPMeOH (500 pg pL-') standard mixture and 50 pL of a 15 MeSOrPCB (211-604 pg pL-l),SMeSOrDDE (595 pg pL-9, and IS (436 pg pL-l) standard mixture (Table 1) prior to a second cycle of the GPC Figure 1,position B) . Contaminantfree lipid extract weights and MeSOrPCB and MeSOTDDE spiking levels were chosen to simulate coextracted lipid loads and MeSOrmetabolite levels encountered in the analysis of polar bear liver and adipose tis~ue.~,9 PCBs were spiked only to contaminantfree lipid extracts because of relatively high natural PCB levels in the tissue substrates. A; Litzen, K; Nyland, K; ReutergArdh, L;Sellstram, U.; Uvemo, U.-B.; Wahlberg, C.; Wideqvist, U.Fresenius J. Anal. Chem. 1991,340, 439-445.
(21) Jansson, B.; Andemson, R; Aspland, L.; Berg",
Analytical Chemistry, Vol. 67,No. 22, November 15, 1995
4157
Contaminant Enrichment and Cleanup Procedure. The first step in enrichment involved removal of lipids by GPC.22 After rotary evaporation, the lipid extract was diluted to 10 mL with 50:50 dichloromethane/n-hexane. An accurate 10%volumetric portion was removed to determine lipid gravimetrically.8 The remaining volume was readjusted to 10 mL, vortexed, and loaded onto the GPC. The first 140 mL fraction, containing the lipids, was dumped or used for contaminant-free spiking. The second fraction (150 mL), containing CHCs and metabolites, was collected, reduced in volume to -2 mL by rotary evaporation, and eluted through the 33% KOH/silica gel column (1.5 g) with 50 mL of 50:50 dichloromethane/n-hexane. The extract volume was reduced to -1 mL and chromatographed on Florisil (8.0 g). The first fraction (75 mL of 50:50 dichloromethane/n-hexane) contained all routinely determined CHCs, including PCBs and TCPMeOH, substantially the same as the sum of the three fractions in the method outlined in Norstrom et al.22,23The second fraction (80 mL of 7:93 methanol/dichloromethane) contained the aryl methyl sulfones. This fraction was rotary evaporated at 30 "C under reduced pressure until just dry. The residue was redissolved in n-hexane and chromatographed on a 2.3% HzO deactivated basic alumina column (3.0 g). The first 10 mL of 50: 50 dichloromethaneln-hexane was discarded, and the second 40 mL fraction, containing aryl methyl sulfones, was collected using the same solvent mixture. The aryl methyl sulfone and CHC fractions were adjusted to 1-2 mL by rotary evaporation and quantitatively transferred to 5 mL, acid-washed borosilicate glass vials with 3 x 0.5 mL n-hexane washes and -50 pL of 2,2,4 trimethylpentane. The volume was reduced to -100 pL by a gentle stream of Nz gas (99.99% purified) at ambient room temperature. The vial sides were washed down with -100 pL of 2,2,4trimethylpentane,and the volume was adjusted very slowly at ambient temperature by Nzassisted solvent evaporation to 50 pL, followed by brief vortexing and transfer to a GC injection vial. Instrumental Analysis. GC-ECD. Capillary GC-ECD was performed on a Hewlett-Packard (Palo Alto, CA) 5890 equipped with a 63NiECD detector and HP 7673A automatic injector. The GC was fitted with a fused silica, Rtx-5 capillary column (Restek Corp., Bellefonte, PA; 30 m, 0.25 mm id., 0.1 pm film thickness, cross-linked 95%dimethyl-5% diphenyl polysiloxane). The carrier gas and ECD makeup gas were helium and 5% methane-95% argon, respectively. All injections were 2 pL in volume and made in the splitless mode. GC conditions and temperature ramping were optimized for MeSOzPCB and -DDE separation.8 The injector and detector temperatures were set at 270 "C and 330 "C, respectively, and the GC oven temperature program was as follows: initial temperature held at 100 "C for 3 min, 20 "C/min to 220 "C, and 3 "C/min to 280 "C. For PCBs and TCPMeOH, the injector and detector temperatures were 250 "C and 320 "C, respectively, and the GC oven temperature program was as follows: initial temperature held at 70 "C for 2 min, 5 "C/min to 150 "C, and 3 "C/min to 250 OC. GC-FID. The instrument and capillary column for validation of MeSOzPCB and -DDE standard concentrationswere the same as for GC-ECD. The MeSOrPCB and -DDE and IS injection technique was cold oncolumn using the "four-segment injection" approach:200.5 p L of pure solvent (2,2,4trimethylpentane), 0.5 (22) Norstrom, R J.; Simon, M.; Mulvihill, M. J. Int. /. Enuiron. Anal. Chem. 1986,23,267-287. (23) Norstrom, R. J.; Simon, M.; Muir, D. C. G.; Schweinsburg, R Enuiron. Sci. Technol. 1988,22, 1063-1071.
4158
Analytical Chemistry, Vol. 67, No. 22, November 15, 7995
pL of air, 0.6 pL of sample, and 0.8 pL of air. On-column injection was used to prevent discrimination of high molecular weight compounds. The detector temperature was 350 "C, and the GC oven temperature program was as follows: initial temperature held at 85 "C for 3 min, 20 "C/min to 200 "C, and 2.2 "C/min to 265 "C. The carrier gas was helium. The makeup gas had a total flow rate of 440 mL/min (380 mWmin compressed air, 30 mL/ min helium, and 30 mWmin hydrogen). GC/ECNZ-MS. Total ion current O C ) scanning was from 83 to 550 amu. The (M- and (M 2)- ions were scanned in the selected ion monitoring (SIM) mode, and the sum of their responses was determined for each congener. The ions scanned were m / z 368/370,404/406,438/440,472/474,418/420, and 3941 396 for the tetrachloro- to heptachloreMeSOzPCB isomer groups, IS, and SMeSOrDDE,respectively. GC/ECNI-MS was performed on a HP 5985 mass spectrometer upgraded with an HP 5890 series I1 gas chromatograph coupled with an HP 5988 GC/MS direct interface. The GC column was the same as that used for GCECD and operated in the splitless mode. The camer and reagent gases were helium and methane (0.5 Torr), respectively. The instrument was tuned with perfluorotributylamine (PFTBA) at m / z 312,414, and 595 for optimal conditioning. The emission current was 300 pA, and the electron voltage was 130 eV. The injection port, transfer line, and ion source temperatures were 270 "C, 260 "C, and 200 "C, respectively. The oven temperature program was the same as that used for GC-ECD. A 2 pL volume was used for all splitless injections. Determination of Recoveries by GC-ECD. All analyses were done in triplicate. For the contaminant-free lipid extracts, herring gull egg, and smelt substrates, areas of the MeSOzPCB, SMeSOrDDE, PCB, and TCPMeOH peaks relative to the IS were compared to those in an external standard to determine percent recoveries. The native MeSOzPCB and 3-MeSOTDDEcongener levels in polar bear liver and fat substrates were first determined by GGECD, and 12 of the MeSOaPCB and 3-MeSOTDDE standardswere spiked to a second set of 0.5 g samples at matching levels. A spike of 100pL of 100 pg pL-' IS was also added (Figure 1,position A). The sample size was adjusted so that native MeSOz PCB and -DDE levels in the final fraction were in the 25-500 pg pL-' range.8~~~ The GC-ECD peak area contributions of the analytes of the unspiked replicates (n = 3) were subtracted from the spiked replicates (n = 3) for the tissue substrates. The difference in peak areas was compared to the original MeSOz PCB and -DDE standards to determine percent recoveries for method validation. MeS02-PCB and MeS02-DDEResponse Parameters. A separate mixture of -1 ng pL-' per congener of the 15 MeSOT PCB, SMeSOrDDE, and IS standards used for method validation (T'able 1)was prepared and serially diluted down to -0.01 pgpL-'. The IDLs for ECD, SIM (0- (M + 2)-), and TIC for each congener were deiined as the mass in picograms giving a signalto-noise (S/N) ratio of 3.25 Replicate (n = 7) 0.5 g wet wt equiv of polar bear liver, contaminant-free lipid extract samples were spiked with the standard mixture at concentrations near the (24) Letcher, R J.; Norstrom, R J.; Bergman, A,; Muir, D. C. G. Dioxin '92,
+
+
Orgcnohulogen Compounds; Tampere, Finland, Aug 24-28, 1992; Finnish Institute of Occupational Health: Helsinki, Finland, 1992; Vol. 8, pp 357360. (25) MacDougall, D.; Lal,J.; Amore, F. J.; Langner, R R; Cox, G. V.; McClelland, N. I.; Crosby, D. G.; Phillips, W. F.; Freeman, D. H.; Pojasek, R. P.; Gibbs, W. E.; Severs, R E.; Gordon, G. E.; Keith, L. H. Anal. Chem. 1980,52, 2242-2249.
Table 2. Mean Percent Recoveries for MesopPCB$, &MeSOpDDE, and the Intemd Standard from Tissue Substrates and MeSOpPCBs, 34eSOpDDE, IS, PCBs, and TCPMeOH from Contaminant-Free Lipid Extracts Determined by QC-ECD
hemng gull egg
wet wt (g) % lipid
smelt
polar bear liver
polar bear adipose
0.5 2.7
0.5 65.0
Tissue Substratesn 5.0 10.0
15.0 4.4
Methyl Sulfones mean % recovery 77f5 78f4 83f5 relative mean % recoveryb 88 f 6 104 f 5 108 f 7 Contaminant-Free Lipid Extractsn lipid extract wt (g) 0.18 0.15 0.019 Methyl Sulfones mean % recovery 78f5 75f3 76f4 relative mean % recoveryb 105 f 7 105 f 3 102 f 5 PCBs mean % recovery 75f7 79f1 84f5 relative mean % recoveryb 101 f 9 110 f 3 113 f 7 TCPMeOH relative mean % recoveryb 116 f 2 110 f 3 114 f 2
*
83 3 104 2z 4
16
18
22
20
0.65
24
26
11 ..
80 f 3 102 f 4
*
79 3 100f4
15
/I I
1
99 f 3
fSD amon congeners. See Table 1 for congener structures in each matrix spife. There are 14 dichloro- to octachloro-PCBs in the NRC-CLB-1-D solution. Relative to the IS.
22
18
24
22
20
26
IDL for determination of the corresponding MDLs. The MDLs were defined as three times the significance of variation (SV) calculated using Student's t statistic. The SVs were calculated from the SDs (n = 7) using the appropriate Student t-values according to
SV = ~ D M D L ~ - I , ~ ~ % ) (1) where t(,,-~, 95%) was the t-distribution constant for n - 1 degrees of freedom at the 95%confidence level. The commonly accepted MDL is 3 SV, and the practical quantitation limit (PQL) is 10 SV (i.e., 3.3 MDL).25,26 MRFs for ECD, SIM, and TIC were determined in triplicate at several concentrations from -1000-1700 pg down to the MDLs to determine the linear concentration range. The variance in the MRFs, IDLs, and MDLs were assessed by a oneway analysis of variance (ANOVA) using Quattro Pro 6.0 (Novell Inc., Orem, VT). The sigdicance level of the TIC response variance was 95%for n = 3 replicates at a given concentration and n = 6 concentrations. There were n = 7 different concentrations for ECD and SIM. The MRFs relative to the IS (RMRF) were also calculated. RESULTS
Recovery Studies. The 15 MeSOTPCB standards chosen for method validation (Table 1) represented a majority of the 22 congeners identified so far in b i ~ t a . * ~ ~ The J ~ Jmean ~ ~ ~ percent * recoveries obtained for the analyte standards spiked to tissue substrates and contaminant-freelipid extracts are shown in Table 2. Percent recoveries were consistently greater than -75% regardless of tissue matrix or variation in sample weight. The SDs of the mean recoveries indicated that there was no significant discriminationamong MeSOTPCBs, SMeSOTDDE,IS, PCBs, and TCPMeOH, substrates or sample weights. High method precision (26) Tuner, W. E.;Patterson, D.G., Jr.; Isaacs,S.G.;Alexander, L R Chemosphere 1992,25,793-804.
7
z 12
14
16
18
20
Retention Time (min)
Figure 2. Chromatograms of the aryl methyl sulfone fraction obtained from the column chromatography methodology: (A) GCECD of polar bear fat, (B) GC-ECD of polar bear liver, and (C) GCI ECNI-MS(TIC) of polar bear liver. Peaks numbers refer to the MeS02-PCB and -DDE structures in Table 1. Peaks denoted CI, were identified as MeS02-PCB isomers by GC/ECNI-MS(TIC). The approximate GC elution window for MeS02-PCB isomer groups are denoted in chromatogram A. The shaded peaks were non-MeSO2PCB or -DDE compounds.
was indicated by the low SD of percent recovery of individual congeners, which ranged from 0.3% to 10%but was generally between 1%and 6% (not shown). MeS02-PCB and -DDE Congener Identity and Method Cleanup ElXciency. GGECD chromatograms of the aryl methyl sulfone fraction of unspiked polar bear fat and liver samples are shown in Figure 2A,B, and the GC/ECNI-MS(lIC) mass chromatogram of unspiked polar bear liver is shown in Figure 2C. virtually all of the peaks were identified as specific MeSOzPCB congeners by comparison to authentic standards or as isomeric Analytical Chemistv, Vol. 67, No. 22, November 15, 1995
4159
Table 3. Mean In8trumentaf and Method Detection E , the Internal Limits for MeSOrPCB8, ~ M ~ S O Z - D Dand Standard
'I
GC/ECNI-MS GC-ECD
\
(M-SO2CHjtH)-
7 -
280
300
., , , ,
320
IDL (pg)"$c 0.23 (87) (241 MDL (pg)b*c 6.15 (85) [191 SV 2.42
II
300
340
320
340
360
340
360
380
400
420
I 280
300
320
TIC
0.08 (47) [341 0.55 (34) [391
0.99 (37) [41 11.25 (59) [311 4.44
0.21
IDLs of the sulfone standards used for method validation (Table l),n = 3 replicates analysis for each congener. For MDLs, the Same standards used for IDL determination were spiked at concentration levels near the IDLs to n CHC-free polar bear liver lipid extracts (n = 6 for ECD and TIC, n = 7 for SIM). The % CV contributions of amongcongener variance (parentheses) and among-replicatevariance [brackets] of total variance were determined by a one-way ANOVA. Mean significance of variation (eq 1) of congener MDLs.
,.I,*
(M-CI+H)'
280
SIM
380
400
420
m/z ratio
Figure 3. Representative ECNl mass spectra for (A) 3-MeS02CB101, (B)4-MeS02-CB101, and (C) 3-MeS02-DDE selected from the mass chromatogram of the aryl methyl sulfone fraction from polar bear liver (Figure 2). The molecular and major fragment ions are labeled. A MeSO2-CI5CBimpurity coeluted with 3-MeS02-DDE and is labeled.
MeSOTPCBs from the GC/ECNI-MS mass spectra. Native levels of total MeSOrPCBs in polar bear liver and fat samples were -2000 and -400 ng g-1 on a lipid weight basis, respectively, consistent with results found p r e v i o u ~ l y . ~ * ~ ~ The MeSOTPCB and -DDE congener pattern in herring gull egg (not shown) was similar to that in polar bear adipose and liver (Figure 2). Herring gull egg had relatively high native levels of total MeSOrPCBs, estimated to be -150 ng g-' on a lipid weight basis. Despite substantial levels of PCBs and other CHCs, aryl methyl sulfones were not detectable in smelt. The aryl methyl sulfone elution window for smelt and hemng gull egg substrates was also essentially free of interferences for both GC-ECD and TIC (not shown). GC/ECNI-MS mass spectra of native 3- and 4MeS0~CB101 and 3-MeSOrDDE (peaks 7,9, and 10, respectively, Figure 2C) in polar bear liver are shown in Figure 3A-C, respectively. All ions arose from molecular and fragment anions of MeS02-PCBs 4160 Analytical Chemistry, Vol. 67, No. 22, November 15, 1995
and -DDE. Similarly clean mass spectra were found in the aryl methyl sulfone fraction of all four tissue substrates. MeS02-PCB and MeS02-DDE Response Parameters. The mean MDL was generally subpicogram for SIM and in the low picogram range for ECD and TIC (Table 3). The mean MDL to mean SV ratios were -2.5 for the three techniques. A ratio of -3 is acceptable for reliable, low mass quantitation. The ratio of mean MDL to mean IDL was seven for SIM, 11for TIC, to 27 for ECD. ANOVA of the MDL and IDL data was performed to determine the relative significance of amongcongener and amongreplicate variance (Table 3). The among-congenervariance was larger than among-replicate variance, especially for TIC. Trends in MDL and IDL values with respect to MeSOTPCB and -DDE structure were not obvious for any technique. The mean PQLs among congeners, defined as 10 SV, were 24.2 f 4.6 pg for ECD, 2.1 f 0.9 pg for SIM, and 44.4 f 17.1 pg for TIC. Mean MRFs were relatively undeviating among MeSOTPCB congeners for all three techniques, especially SIM (Table 4). There was less than a factor of 2 difference between the highest and lowest mean MRFs for each technique. There was not a pronounced tendency for mean MRFs to increase with increasing degree of chlorination from four to seven chlorines per molecule. No consistent pattern in MRF values between 3- and 4MeSOr PCB pairs could be discerned because differences were similar to the SDs for each congener and within the expected accuracy of the standard concentrations in most cases. By far the largest difference occurred between 3- and 4MeS0~CB101.The mean MRF for 3-MeSOz-DDE was similar to those of the MeSOrPCBs. The constancy of MRFs for all congeners using the three detection techniques, over a concentration range from low picogram levels up to 1.7 ng, was similar to that represented by the four MeSOT PCBs in Figure 4. The exact threshold congener masses at which serious deviation from linearity began were not determined, but they were less than the PQL. ANOVA of the MRF data was performed to determine the relative signi6cance of among-mass and among-replicate variance, which are given as % CV in Table 4. The mean F-ratios among congeners for ECD (18.68 f 14.68) and TIC (55.67 rt 38.17) far exceeded the critical value (2.51)27for the null hypotheses that amongmass variance was the same as among-replicate variance. Thus, the among-mass variance was signi6cantly larger (e.g., the (27) Pearson, E. S.; Hartley, H. 0. Biometrika Tables for Statisticam, 2nd ed.; Cambridge University Press: Cambridge, UK, 1958; Val. 1.
Table 4. Mean Molar Response Factors and MRFs Relative to the IS for MeSO#CB and 3MeSOrDDE.
GC/ECNI-MS GC-ECD
SIM (M)-
+ (M+ 2)-
TIC ~
C1
congener
C4
5MeSOpCB49 4-MeSOrCB49 4-MeSOpCB64 5MeSOpCB101 4-MeSOrCB101 3-MeSOrCB87 4-MeSOrCB87 5MeSOpCB149 4-MeSOrCB149 5MeSOrCB132 4-MeSOrCB132 3-MeSOrCB141 4MeSOrCB141 3-MeSOrCB174 4-MeSOrCB174 mean % CV mean total % CV SMeSOrDDE IS
CIS
CIS
c17
RMRFC
MRFb 0.46 0.38 0.44 0.39 0.56 0.62 0.66 0.63 0.57 0.67 0.58 0.65 0.64 0.75 0.59
(8.6) (9.2) (8.5) (7.4) (8.3) (9.1) (9.5) (6.2) (8.7) (6.5) (5.5) (8.6) (6.7) (6.0) (4.6) (7.6) (9.9) 0.70 (9.4) 0.69 (5.6)
L2.41 12.21 l2.21 l2.01 [2.21 [2.21 l2.21 [2.11 L2.61 [2.4] [2.2] [2.41 [2.31 [2.41 [2.61 [2.31 [1.91 [2.31
0.66 0.56 0.63 0.57 0.81 0.89 0.95 0.91 0.82 0.96 0.84 0.95 0.93 1.09 0.85
(Q.45) (5.4) (4.8) (1.8) (2.5) (4.5) (4.2) (3.3) (4.9) (1.0)
(2.4) (2.1) (1.1) (2.8) (2.4)
(3.2) 1.02 (7.8)
RMRFC
MRFb 1.33 1.62 1.33 0.93 0.73 1.24 1.29 1.34 1.45 1.45 1.35 1.39 1.62 1.20 1.06
(21.5) (19.4) (29.7) (27.2) (17.0) (28.0) (18.7) (21.5) (21.5) (26.1) (29.0) (29.0) (28.4) (25.6) (16.0) (23.9) (39.8) 1.51 (24.1) 1.20 (13.0)
[13.21 [13.81 [15.6] [16.51 [16.2] f22.71 15.81 [13.4] [15.0] [14.3] [18.0] [27.11 [14.21 [15.3] [17.4] [15.9] [16.5] [5.4]
1.12 (19.6) 1.35 (14.8) 1.11 (22.5) 0.81 (21.0) 0.61 (18.0) 1.04 (24.0) 1.08 (7.4) 1.12 (17.8) 1.22 (21.3) 1.22 (22.1) 1.13 (22.1) 1.17 (28.2) 1.36 (22.0) 1.01 (21.8) 0.89 (19.1) (20.1) 1.26 (20.6)
MRFb 0.82 0.81 0.75 0.65 0.86 0.96 1.18 1.06 0.99 1.11 0.95 1.16 1.15 1.14 0.96
(9.0) (9.9) (9.7) (9.0) (8.1) (12.1) (9.6) (11.5) (10.8) (9.6) (13.0) (12.4) (13.1) (14.1) (15.0) (11.1) (13.8) 1.06 (12.6) 1.06 (13.4)
RMRFC [2.2] 14.11 [3.01 [2.0] f2.71 [2.3] f3.41 f2.51 [3.0] [1.7] [2.0] [3.9] [2.01 [3.6] [2.81 [2.71 [1.8] 13.31
0.78 0.77 0.71 0.62 0.82 0.90 1.12 1.01 0.93 1.06 0.90 1.10 1.09 1.08 0.91
(8.9) (10.4) (9.8) (8.1) (9.8) (10.0) (10.7) (8.9) (6.4) (9.4) (12.2) (7.3) (9.2) (8.3) (5.5)
(9.0) 1.02 (17.6)
a MRF x 1OI8 (peak area counts/mol) for analysis at i mass levels from -20 to 1700 pg for ECD and TIC and from -20 to 600 pg for SIM: i 7 for GC-ECD and GC/ECNI-MS(S1M) and i = 6 for GC/ECNI-MS(TIC). At each i mass level, n = 3 replicates. The % CV contributions of among-massvariance (parentheses) and among-replicatevariance [brackets]of total variance were determined by a one-wayANOVA. See Results section for the F-statistics. The % CV (parentheses) is for the total variance.
=
mean % CV for ECD was 7.6) than among-replicatevariance (e.g., the mean % CV for ECD was 2.3) for these techniques Vable 4). The mean F-ratio among congeners (3.04 f 2.24) and F-critical value (2.85)27were very close for SIM, indicating that amongreplicate variance was closer to, but still not significantly higher than, among-mass variance for this technique. Mean MRFs for SIM also had the greatest total variance. It therefore appears that SIM is an inherently less reliable technique than ECD and TIC. The mean MRFs relative to the IS (RMRFs) ranged from 0.6 to 1.4 for all three techniques (Table 4). The total % CV of the mean RMRFs were less than the total % CV of the corresponding mean MRFs. The deviation from linearity of MRFs for ECD and TIC at low concentrations (Figure 4A,B) was improved slightly after conversion to RMRFs. In contrast, RMRFs for SIM declined more at low concentrations than did MRFs (Figure 4C). Fluctuations in the congener MRFs for SIM over the entire concentration range were not reduced after conversion to RMRFs. DISCUSSION
Extraction and Cleanup/Qualitaiive Analysis. The present column chromatography-based extraction and cleanup method (Figure 1) is simple and nondestructive and is applicable to determination of CHCs in a wide range of animal tissue substrates. Although the method was not validated for all CHC compound classes normally determined, the normal range in polarity (elution order from Florisil and alumina columns) was spanned by PCBs and TCPMeOH. The method was extended to the even more polar aryl methyl sulfones. This method is therefore suitable for multiresidue, routine determination of CHCs, including aryl methyl sulfones, in animal tissues. The method has advantages over those previously used for aryl methyl sulfones, which included liquid-liquid partitioning steps2,9-13 or sap~nitication.~~J~ There is a risk of analyte loss or fraction contamination using liquidliquid partitioning. Strong acid and base conditions destroy some
CHCs, rendering these methods unsuitable for multiresidue analysis. For example, higher chlorinated dibenzc-pdioxins are base and the oxygenated organochlorines, oxychlordane and dieldrin, are strong acid labile.29 Most coextractingbiogenic material was separated from CHCs by GPC. Under the same operating conditions, GPC has been shown to remove -99% of long-chain lipids and about half of the carotenoid pigments from chicken egg y o k Z 2 Residual acidic biogenic material remaining in the CHC fraction after GPC is also retained. In a separate experiment, we were able to retain approximately half the lipid extract weight from polar bear adipose tissue on the 33% (w/w) KOH-impregnated silica gel column. The aryl methyl sulfones were cleanly separated from other CHCs, including TCPMeOH, by chromatography on 1.2%waterdeactivated Florisil. The aryl methyl sulfones require high-polarity solvents for quantitative elution from Florisil. Neat methanol has been used for elution of MeSOzPCBs and -DDE with no further cleanup, although s i m c a n t levels of residual interferences were observed in the aryl methyl sulfone f r a ~ t i o n . ' ~The , ~ ~presence of long-chain lipids was unlikely since lipid extracts had been subject to GPC prior to Florisil. However, residual interferences from polar compounds with smaller molecular diameters, such as retinols and sterols, may have coeluted in the aryl methyl sulfone fraction. These compounds are sufficiently electrophilicas a result of conjugation or oxygenation to be ECD and ECNI-MS respon~ i v e .In~ the ~ present study, a 7:93 methanol/dichloromethane solvent mixture eluted MeSOzPCBs and -DDE while retaining residual colored biogenic coextractants. Color retention was (28) Lamparski, L. L.; Nestrick, T.J.; Stehl, R. H. Anal. Chem. 1979,51,14531458. (29)Smith, L.M.;Stalling, D. L.; Johnson, J. L. Anal. Chem. 1984,56, 18301842. (30)Buser, H.-R; Zook, D. R ; Rappe, C. Anal. Chem. 1992,64, 1176-1183. (31) Gum, M. I.; James, A T. Lipid Biochemistry: An Introduction, 3rd ed.; Chapman and Hall: London, 1980; Chapters 1 and 3.
Analytical Chemistry, Vol. 67, No. 22, November 15, 1995
4161
1.6 1.2
3
0.8 0.4
0 1.6
1
I
!
1.6 1.2 0.8 0.4 I
0
:
0
:
:
:
200
.
:
400
600
Congener Mass @g) Figure 4. Mean molar response factor (MRF, peak area counts/ mol of compound x 1Ol8, n = 3 per point) profile of representative MeS02-PCBs over a concentration range commonly used for determination of CHCs8,9,12,13,33 by (A) GC-ECD, (8)GC/ECNI-MS(TIC), and (C) GC/ECNI-MS(SIM). Q3-MeSOz-CB49; v, 3-MeS02-CB101; 0, 3-MeSOz-CB149; and A, 3-MeS02-CB174. Error bars on individual points were not shown for clarity; see Table 4 for response variation contributions of the mean MRFs. The vertical line denotes the approximate low mass limit of MRF linearity.
especially obvious for polar bear liver. Some interferences were observed in the ECD and TIC chromatograms after Florid, but these mostly eluted prior to the aryl methyl sulfone elution window. Remaining interferences in the aryl methyl sulfone window were removed by the 2.3%HzO-deactivated basic alumina column, giving clean chromatograms for all three techniques (Figure 2). Five or fewer minor peaks in the aryl methyl sulfone elution windows of polar bear adipose (Figure 2A) and polar bear liver (Figure 2B,C) were not MeSOrPCB or -DDE compounds. The lack of interfering ions in the ECNI mass spectra for MeSOT PCBs and -DDEs from polar bear liver (Figure 3) also emphasized the absence of non-aryl sulfone compounds. Furthermore, the (M)-/(M 2)- ion abundance ratios of -50:100 and 60100 for 3- and 4MeSOzCB101, respectively (Figure 3A,B), resembled the ideal 61.5100 ratio for pentachloro-MeSOTPCBs. The recently reported bis(khloropheny1) sulfone (BCB) in fish from the Baltic Sea32was found by ECNI-MS in the herring gull egg sample. BCPS eluted in the aryl methyl sulfone window among the hexachloro-MeSOTPCB congeners. It is therefore possible that the present method is suitable for determination of diary1 sulfones. ECD and ECNI-MSQuantitation. Internal standard correction improved accuracy of MeSOTPCB/DDE analysis but overcompensated for PCB and TCPMeOH recovery in smelt and
+
(32) Olsson, A.: Bergman, k Ambio 1995, 24, 119-123.
4162
Analytical Chemistty, Vol. 67, No. 22, November 15, 7995
polar bear liver. Approximately 10%of the losses in the present study were subsequently shown to have occurred because of incomplete sample loading in the GPC step. Recoveries of all analytes from polar bear adipose and liver tissue have now been improved to -90% by optimizing GPC operating Good recoveries have also been obtained for standards spiked to other matrices and isolated by liquid-liquid partitioningbased methodologies. For example, -90% recovery was achieved for five dichloro- to hexachloro-substituted MeSOrPCBs spiked to bovine fat14and I4C-labeledMeSOTDDE spiked to rat tissues? However, routine analysis using these methods has not been thoroughly demonstrated. Precision of the whole analytical method was best demonstrated by the low variance in recoveries of MeSOzPCB and -DDE from spiked polar bear tissues. The recoveries were based on the difference between determination of native levels and spiked levels of the same compounds. The variance in the recoveries was therefore a compound variance of both spiked and unspiked determinations. In spite of this, the SD of replicate analysis of individual compounds ranged from 2%to lo%,with a mean of 6 f 2%,well within an acceptable range for environmental analyses. The best overall precision for MeSOTPCB and -DDE determination was obtained in the analysis of smelt. The mean SD of replicate determinationswas 2.0 2~ 1.1%. Precision of determination of PCBs and TCPMeOH in contaminant-free lipid extracts was also excellent. Only one out of 60 triplicate recovery determinationsfor the PCBs and TCPMeOH had a SD >7%. The SD of the mean recovery among MeSOTPCB and PCB congeners was
