Sakrapport till Naturvårdsverkets Miljöövervakning: Kalibrering av ny analysmetod: Polyklorerade dibenzo-p-dioxiner (PCDD) och dibenzofuraner (PCDF) och polyklorerade bifenyler (PCBer) i modersmjölk. Avtalsnummer: 215 941 Utförare: Livsmedelsverket Programområde: Hälsorelaterad miljöövervakning Delprogram: Specialprojekt - utvecklingsprojekt Undersökningar/uppdrag: Metodjämförelse dioxinanalyser i modersmjölk Marie Aune, Ulrika Fridén, Sanna Lignell Livsmedelsverket 212-1-1 1
SAMMANFATTNING I denna rapport presenteras resultaten från en metodjämförelse som gjorts för analys av polyklorerade dibensodioxiner (PCDD), polyklorerade dibensofuraner (PCDF) och polyklorerade bifenyler (PCB) i modersmjölk. Prover, som tidigare har analyserats inom Livsmedelsverkets (SLV) tidstrendsstudie (för modersmjölk), har analyserats om med SLVs nya metod baserad på gaskromatografi kopplad till högupplösande masspektrometri. Kalibreringsstudien visar jämförbara resultat mellan den nya SLV metoden och de tidigare använda metoderna för summan av de toxiska ekvivalenterna ( TEQ) samt för de kongenerna som bidrar mest till TEQ (2,3,7,8-TCDD, 1,2,3,7,8-PeCDD, 2,3,4,7,8-PeCDF, CB 126, CB 169, CB 118 and CB 156). Den nya SLV-metoden ger något högre halter av indikator- och mono-orto-pcber än de tidigare använda metoderna, dessa kongener ger dock endast ett litet eller inget bidrag till TEQ. Framtida tidstrender för PCB och PCDD/F i modersmjölk kommer att baseras på data från både de tidigare och den nya SLV-metoden. Om tidstrenderna påverkas av bytet av analysmetod kan en justering av halterna av indikator- och mono-orto-pcber bli nödvändig enligt resultaten i denna kalibreringsstudie. 2
Report to the Swedish Environmental Protection Agency, 212-1-1 Marie Aune, Ulrika Fridén and Sanna Lignell Analysis of polychlorinated dibenzo-p-dioxins (PCDDs) and dibenzofurans (PCDFs) and polychlorinated biphenyls (PCBs in human milk a calibration study. INTRODUCTION Among the Swedish human population, food is the major source of exposure to persistent organic pollutants (POPs), such as polychlorinated dibenzo-p-dioxins (PCDDs), polychlorinated dibenzofurans (PCDFs), and polychlorinated biphenyls (PCBs). These compounds are lipophilic and accumulate in the lipid compartment of the human body. The POP levels in body lipids therefore reflect the long-term exposure of the individual. Exposure estimation is an important part of risk assessment of POPs in food. Due to the relatively high lipid content, human milk is a good human matrix for analysis of POP body burdens at the time of pregnancy and nursing. In order to estimate the body burdens of POPs among pregnant and nursing women, and to estimate the intake of the compounds by breast-feeding infants, the Swedish National Food Agency (NFA) has made recurrent measurements of concentrations of POPs in human milk. The project also establishes if there are temporal trends of POP concentrations in human milk and the temporal trends of POPs between 1996 and 28 have been reported (Glynn et al. 27, Lignell et al. 29a and Lignell et al. 29b). PCDDs and PCDFs (PCDD/Fs) were analysed at the National Institute of Public Health and the Environment (RIVM), the Netherlands, from 1996 to 24, using methods described in Glynn et al. (21). In 26, the analyses were performed by the Department of Chemistry, Environmental Chemistry, Umeå University (UmU) using a method described in Danielsson et al. (25). Non-ortho PCBs were analysed at RIVM (1996-1999), the NFA (2-24), or at Umeå University (26). Mono-ortho and indicator PCBs have been analysed at the NFA during the whole time period 1996-28. For the methods used at the NFA see references in Lignell et al. (29b). 3
During year 2, a calibration study was performed for non-ortho PCBs, in which 26 samples were analysed at both the NFA and RIVM in order to calibrate the results of the two laboratories (Lignell et. al. 26). The comparison showed no systematic errors. During year 26, a calibration study was performed, comparing the PCDD/F, CB 126 and CB 169 results in 1 samples analysed both at RIVM and UmU (Glynn et. al. 27). Results of non-ortho analyses performed at the NFA and UmU were also compared. The comparison showed that UmU in most cases reported somewhat higher median levels of PCDD/Fs and non-ortho PCBs than RIVM and the NFA. These differences, however, did not significantly affect the results of the temporal trends that have been investigates earlier. In order to continue the study of temporal trends, samples of human milk from year 28 and 21 will be analysed for PCCD/Fs and PCBs by the NFA using a new method. The new method is based on gas chromatography coupled to high resolution mass spectrometry (GC- HRMS), which is the same technique as has previously been used by RIVM and UmU. In this calibration study, results obtained with the new NFA method have been compared with the results obtained from the two previously used laboratories as well as from the previous NFA methods for PCBs. MATERIAL AND METHODS Human milk samples Table 1 shows the selection of human milk samples included in the calibration study (N=2), and where they have been analysed previously. All samples have previously been analysed within the temporal trend study. When sufficient amount of sample was available, samples that previously had been analysed by both UmU and RIVM were chosen. Included are also some samples analysed by only one of UmU or RIVM. Method blanks and a human milk control sample were also analysed in parallel to the samples. Analytical methods Table 2 shows the compound groups and the individual congeners that were analysed using the new NFA method in the human milk samples. Some of the analytes in Table 2 have not been analysed previously (CB 123 and CB 189 in all samples, and CB 81, CB 114 and CB 157 in some samples). 4
Table 1. Summary of substance groups previously analysed, and where the previous analysis has been performed for the samples included in the calibration study. Sample Code PCDD/Fs Non-ortho PCBs Indicator and mono-ortho PCBs HF71 RIVM NFA NFA HF718 RIVM a RIVM NFA HF817 RIVM a RIVM NFA HF91 RIVM RIVM NFA HF914 RIVM RIVM NFA HF916 UmU RIVM NFA HF242 UmU UmU / NFA NFA HF2423 UmU UmU / NFA NFA HF2447 UmU UmU / NFA NFA H696 UmU UmU NFA H699 UmU UmU NFA H611 UmU a UmU NFA H618 UmU UmU NFA H619 UmU UmU NFA H611 UmU UmU NFA H6119 UmU a UmU NFA H6126 UmU UmU NFA H6129 UmU a UmU NFA H6132 UmU UmU NFA H6134 UmU a UmU NFA a) No new NFA results for comparison Table 2. Summary of compound groups and individual congeners included in the calibration study on PCBs and PCDD/Fs in human milk samples. Compound group Congeners Indicator PCBs 28, 52, 11, 138, 153, 18 Mono-ortho PCBs 15, 114, 118, 123, 156, 157, 167, 189 Non-ortho PCBs 77, 81, 126, 169 PCDDs 2,3,7,8-TCDD 1,2,3,7,8-PeCDD 1,2,3,4,7,8-HxCDD 1,2,3,6,7,8-HxCDD 1,2,3,7,8,9-HxCDD 1,2,3,4,6,7,8-HpCDD OCDD PCDFs 2,3,7,8-TCDF 1,2,3,7,8-PeCDF 2,3,4,7,8-PeCDF 1,2,3,4,7,8-HxCDF 1,2,3,6,7,8-HxCDF 2,3,4,6,7,8-HxCDF 1,2,3,7,8,9-HxCDF 1,2,3,4,6,7,8-HpCDF 1,2,3,4,7,8,9-HpCDF OCDF 5
The extraction and clean-up steps in the new NFA method are based on different published methods, application notes as well as information from experienced colleagues. Briefly, the milk samples (1 g) were treated with a solution of potassium oxalate (35% in water) and ethanol and the analytes were then extracted with diethyl ether and n-pentane (1:2.4 v/v). The extract was washed with a solution of sodium sulphate (2% in water), dried using dried sodium sulphate, followed by evaporation of the solvent and gravimetric lipid weight determination. Lipid removal was performed using a multi-layer silica gel dioxin column (35cm, Ø 3mm i.d., Supelco) eluted with n-hexane. Isolation of PCBs and PCCD/Fs from interfering substances and separation into different fractions was performed in two steps. Firstly, a Florisil column (1-2 mesh, Acros Organics) eluted with.2% toluene in n-hexane and toluene. Secondly, a HyperCarb column (1x4.6mm, 7μm, 25 A, Thermo Scientific) eluted with n-hexan/dichloromethane (2:1) in forward direction and toluene in reverse direction. This resulted in three fractions containing (i) the indicator and mono-ortho PCBs, (ii) the nonortho PCBs, and (iii) the PCCD/Fs. The final determination was performed using a GC-HRMS (Agilent Technologies 798 GC and an AutoSpec Premier, Waters) with the isotopic dilution standard method ( 13 C- labelled surrogate standards for all congeners). The indicator PCBs were injected on a HT8 column with a split/splittless injector in splittless mode. The mono-ortho PCBs, non-ortho PCBs and PCDD/Fs were injected on a DB5-MS column with a PTV injector in solvent vent mode. The HRMS was operated in EI mode, using single ion monitoring (SIM) at the resolution of 1. Previously, PCDD/Fs were analysed by GC-HRMS at RIVM and UmU. The non-ortho PCBs were previously analysed by GC-HRMS at RIVM and UmU or by gas chromatography coupled to low resolution mass spectrometry (GC-LRMS) at the NFA. The mono-ortho and indicator PCBs were previously analysed by gas chromatography coupled to electron capture detection (GC-ECD) at the NFA. RESULTS AND DISCUSSION Linear regression analysis was used to investigate associations between results from previous and new methods for analysis of PCBs and PCDD/Fs in milk. The intercept between previous and new results was set to zero in the regression models. Figure 1 shows the results of the 6
linear regression analyses. The results are presented as toxic equivalent (TEQ) values for nonortho PCBs, mono-ortho PCBs, PCDD/Fs and the total TEQ, calculated using 25 WHO toxic equivalent factors (TEF) (Van den Berg et al. 25). Total TEQ corresponds to the sum of PCDD/Fs and dioxin-like PCBs, where the dioxin-like PCBs is the sum of the non- and mono-ortho PCBs. Data for some PCB congeners are missing for the previous results (i.e. CB 123 and CB 189 in all samples, and CB 81, CB 114 and CB 157 in some samples); however, these congeners have a minor contribution to the TEQ values. The TEQ for the monoortho PCBs were somewhat higher for the new NFA results than for previous results (the slope of the regression line, i.e. the regression coefficient, (k) =.8), while the regression coefficients for the other TEQs were close to 1. The missing results for some congeners in the previous analysis cannot explain the difference. All associations were strong, with R 2 - values ranging from.85 to.96. A summary of the total TEQs for individual samples and the ratios between the different results are shown in Table 3. TEQ (non-ortho PCBs) TEQ (mono-ortho PCBs) 12 2 8 1 4 y = 1.7x R² =.85 4 8 12 y =.8x R² =.96 1 2 2 TEQ (PCDD/Fs ) 3 Total TEQ 15 1 5 y = 1.2x R² =.94 y = 1.4x R² =.98 5 1 15 2 1 2 3 Figure 1. Associations between previous and new NFA TEQ for non-ortho PCBS, mono-ortho PCBs, PCDD/Fs and total TEQ (sum of PCDD/Fs, mono-, and non-ortho PCBs) in human milk samples. Data of some PCB congeners are missing in previous results; i.e. CB 123 and CB 189 in all samples, and CB 81, CB 114 and CB 157 in some samples. The intercepts are set to zero in the regression analyses, TEQ values are based on 25 WHO TEFs (Van den Berg et. al., 25). 2 1 7
Table 3. Comparison of previous and new NFA results of total TEQ (PCDD/Fs, mono- and non-ortho PCBs) in human milk samples. TEQ values are based on 25 WHO TEFs (Van den Berg et al. 25). Sample Code Previous results New NFA results Previous/ New NFA Total TEQ [pg/g lw] Total TEQ [pg/g lw] HF71 25.9 a 25.6 1.1 HF91 16.7 b 16. 1.4 HF914 17. b 15.5 1.1 HF916 1.7 c 12.3.87 HF242 11. d 1.7 1.3 HF2423 8.41 d 8.59.98 HF2447 6.31 d 5.97 1.6 H696 7.51 d 6.71 1.12 H699 8.87 d 8.98.99 H618 6.98 d 6.39 1.9 H619 7.39 d 6.44 1.15 H611 11.3 d 9.96 1.14 H6126 8.81 d 8.13 1.9 H6132 9.37 d 8.41 1.11 Mean 11.2 1.7 1.6 Stdev 5.4 5.4.8 a) CB 81, 114, 123, 157 and 189 are not included. b) CB 114, 123, 157 and 189 are not included. c) CB 123 and CB 189 are not included. Concentrations below LOQ were set to 1/2 LOQ d) CB 123 and 189 are not included. 2 1 6 2,3,7,8-TCDD y =.95x R² =.87 1 2 1,2,3,7,8-PeCDD 1 8 6 4 2 2 CB 126 CB 118 y = 1.13x R² =.83 2 4 6 8 4 2 y = 1.3x R² =.84 2 4 6 15 1 5 y = 1.x R² =.97 5 1 15 2 Figure 2. Associations between previous and new NFA concentrations for the individual congeners 2,3,7,8-TCDD, 1,2,3,7,8-PeCDD, CB 126, and CB 118 in human milk samples. The intercepts were set to zero in the regression analyses. 8
Previuos results [ng/g lw] Previuos results [ng/g lw] 2,3,7,8-TCDD, 1,2,3,7,8-PeCDD, and 2,3,4,7,8-PeCDF are the PCDD/F congeners, and CB 126, CB 169, CB 118, and CB 156 the PCB congeners that give the largest contribution to TEQs. Figure 2 shows the strong associations between previous and new NFA concentrations for four of the most important individual congeners. A complete summary of median concentrations, min and max ranges and linear regression coefficients for all individual congeners analysed using the new NFA method is shown in Table 4. Figure 3 shows the association between previous and new NFA concentrations of the most abundant indicator PCB congeners, CB 153 and the indicator PCBs. The associations are strong, but the new NFA method gives somewhat higher concentrations than the previously used method. Slightly lower lipid contents have been indicated using the new NFA method compared to the previous NFA method. This could possibly explain some of the differences between the concentrations of indicator and mono-ortho PCBs obtained from the new and the previous NFA methods. 2 CB 153 4 Indicator PCBs 15 3 1 5 y =.85x R² =.95 5 1 15 2 25 New NFA results [ng/g lw] 2 1 y =.79x R² =.95 1 2 3 4 5 New NFA results [ng/g lw] Figure 3. Associations between previous and new NFA concentrations for the individual congener CB 153 and indicator PCBs in human milk samples. The intercepts were set to zero in the regression analyses. 9
Table 4. Comparisons of PCB and PCDD/F median concentrations (min-max in brackets) in human milk samples analysed using previous and new NFA methods. k is the regression coefficient and R 2 is the coefficient of determination, both obtained from linear regression analysis (with the intercept set to zero). Congener N Previous New NFA k R 2 Indicator PCBs [ng/g lw]: CB 28 2 1.71 (.555-13.8) 2.2 (.686-16.2).83.99 CB 52 2.2 (.12-.48) a.175 (.74-.69) - - CB 11 2.2 (.12-1.6) a.545 (.197-.983) - - CB 138 2 18. (11.1-93.5) 25.8 (15.1-114).72.96 CB 153 2 33.9 (19.2-186) 43.8 (24.4-187).85.95 CB 18 2 16. (9.21-83.8) 2.5 (12.7-92.5).81.97 ind. PCBs 2 73.2 (4.7-366) a 97.2 (56.9-397).79.95 Mono-ortho PCBs [pg/g lw]: CB 15 2 995 (3-339) 145 (629-396).72.77 CB 114 2 3 (145-52) a 341 (171-141) - - CB 118 2 745 (321-176) 695 (371-188) 1..97 CB 123 2 na 59.9 (28.9-124) - - CB 156 2 298 (15-122) 367 (196-173).74.94 CB 157 2 491 (15-118) a 678 (347-36).81.73 CB 167 2 891 (3-29) a 114 (657-343).79.92 CB 189 2 na 321 (94.8-155) - - mono-ortho PCBs 2 129 (638-354) a 151 (87-494).78.96 Non-ortho PCBs [pg/g lw]: CB 77 2 2.6 (1.2-17.) 5.4 (1.7-17.9) b.84.63 CB 81 2 1.2 (.54-5.4) 1.2 (.476-5.45) 1..95 CB 126 2 34. (18.-92.9) 28.2 (16.9-64.9) 1.1.83 CB 169 2 17.7 (1.-43.8) 22.9 (1.4-69.6).8.64 non-ortho PCBs 2 59. (3.-144) 56.4 (36.2-14).98.84 mono- and non-ortho PCBs 2 129 (64-355) a 152 (874-496) b.78.96 PCDDs [pg/g lw]: 2,3,7,8-TCDD 14.67 (.41-1.81).779 (.46-1.94).95.87 1,2,3,7,8-PeCDD 14 1.9 (1.1-4.56) 1.76 (1.11-5.4) 1..84 1,2,3,4,7,8-HxCDD 14.76 (.56-2.3).71 (.379-2.4) 1.2.72 1,2,3,6,7,8-HxCDD 14 6.5 (3.5-19.3) 5.83 (3.47-22.5).91.96 1,2,3,7,8,9-HxCDD 14 1.3 (.79-3.5).79 (.54-2.92) 1.3.86 1,2,3,4,6,7,8-HpCDD 14 9.25 (4.5-25.) 8.66 (3.56-25.2) 1..95 OCDD 14 55. (23.-127) 54. (24.4-123) 1..84 PCDDs 14 73.5 (34.8-182) 67.9 (33.7-183) 1..86 PCDDs [pg/g lw]: 2,3,7,8-TCDF 14.44 (.17-.86).314 (.78-.53) 1.3.73 1,2,3,7,8-PeCDF 14.235 (.19-.65).175 (.95-.477) 1.4.69 2,3,4,7,8-PeCDF 14 4.4 (3.-15.6) 4.412 (2.92-15.7) 1..99 1,2,3,4,7,8-HxCDF 14 1.35 (.9-2.4) 1.5 (.137-1.95) 1.2.75 1,2,3,6,7,8-HxCDF 14 1.25 (.78-1.79) 1.14 (.76-1.93) 1.1.64 2,3,4,6,7,8-HxCDF 14.65 (.34-1.2).528 (.313-1.4) 1.2.71 1,2,3,7,8,9-HxCDF 14.355 (.4-.53) a.25 (-.284) b - - 1,2,3,4,6,7,8-HpCDF 14 1.41 (.68-17.) 1.38 (.8-19.8).87.99 1,2,3,4,7,8,9-HpCDF 14.43 a (.2-.1) a.79 (-1.78) b - - OCDF 14.17 a (.47-1.2) a.2 (-1.81) b - - PCDFs 14 11.2 (8.6-26.3) a 11.9 (6.39-28.9) b 1..86 PCDD/Fs 14 85. (42.9-25) a 78.8 (41.2-26) b 1..86 a) Concentrations below LOQ were set to 1/2 LOQ b) Includes results below LOQ; actual values included in calculated concentrations 1
QA/QC The method blanks for the new NFA method were generally low. Mean concentrations in the blanks were 587, 228, and 4.6 pg /g lw of indicator PCBs, dioxin-like PCBs, and PCDD/Fs, respectively. These blank levels were well below the corresponding median concentrations in the human milk samples 97 2, 15 2, and 78.8 pg/g lw. A human milk control samples was extracted and analysed in parallel to the samples. The mean concentrations were 64 6 (stdev 2 5), 123 (stdev 65), and 31. (stdev 3.8) pg/g lw of indicator PCBs, dioxin-like PCBs, and PCDD/Fs, respectively. CONCLUSIONS The PCB and PCDD/Fs results from the new NFA method are comparable to previous results obtained with other methods. The congeners that contributes most to the total TEQ (2,3,7,8-TCDD, 1,2,3,7,8-PeCDD, 2,3,4,7,8-PeCDF, CB 126, CB 169, CB 118 and CB 156) as well as the TEQ values all show good agreements between previous and new NFA results. The new NFA method gives somewhat higher indicator and mono-ortho PCB concentrations than the previously used method, these congeners, however, give no or a smaller contribution to the total TEQ. In the future, we will investigate temporal trends of PCBs and PCDD/Fs in mother s milk using analytical results obtained with the previous as well as the new method. In these investigations, it is possible that we will have to adjust the concentrations of indicator and mono-ortho PCBs according to the results of this calibration study. ACKNOWLEDGEMENT The Swedish EPA (Environmental Protection Agency) is acknowledged for financial support. Appreciation is expressed to the participating women and to the midwives who assisted in recruitment, interviewing, and sample collection. The laboratory technicians Arpi Bergh, Inga-Lill Gadhasson, Jorma Kuivinen, Lotta Larsson and Elvy Netzel are appreciated for technical assistance. REFERENCES Danielsson C, Wiberg K, Korytar P, Bergek S, Brinkman UA, Haglund P. 25. J. Chromatogr. A. Trace analysis of polychlorinated dibenzo-p-dioxins, dibenzofurans and WHO polychlorinated 11
biphenyls in food using comprehensive two-dimensional gas chromatography with electroncapture detection. 186, 61-7. Glynn A, Atuma S, Aune M, Darnerud PO, Cnattingius S. 21. Environ. Res. 86 (3), 217-228. Glynn A, Aune M, Ankarberg E, Lignell S, Darnerud PO. 27. Report to the Swedish Environmental Protection Agency: Polychlorinated dibenzo-p-dioxins (PCDDs) and dibenzofurans (PCDFs), polychlorinated biphenyls (PCBs), chlorinated pesticides and brominated flame retardants in mother s milk from primiparae women in Uppsala County, Sweden Levels and trends 1996-26. Lignell S, Darnerud PO, Aune M, Törnkvist A, Glynn A. 26. Report to the Swedish Environmental Protection Agency: Polychlorinated dibenzo-p-dioxins (PCDDs) and dibenzofurans (PCDFs), polychlorinated biphenyls (PCBs), chlorinated pesticides and brominated flame retardants in breast milk from primiparae women in Uppsala County, Sweden Levels and trends 1996-24. Lignell S, Glynn A, Törnkvist A, Aune M, Darnerud PO. 29a. Report to the Swedish Environmental Protection Agency: Levels of persistent halogenated organic pollutants (POP) in mother s milk from primaparae women in Uppsala, Sweden 28. Lignell S, Aune M, Darnerud PO, Cnattingius S, Glynn A. 29b. Persistent organochlorine and organobromine compounds in mother's milk from Sweden 1996 26: Compound-specific temporal trends. Environmental Research 19, 76-767. Van den Berg M, Birnbaum LS, Denison M, De Vito M, Farland W, Feeley M, et al. 26. The 25 World Health Organization reevaluation of human and Mammalian toxic equivalency factors for dioxins and dioxin-like compounds. Toxicol Sci 93, 223-241. 12