Dioxin-Like TEQ of Women from the Seveso, Italy Area by ID-HRGC/HRMS and CALUX

www.nature.com/jea

Dioxin-Like TEQ of women from the Seveso, Italy area by ID-HRGC/HRMS and CALUX

MARCELLA WARNER,a BRENDA ESKENAZI,a DONALD G. PATTERSON JR.,b GEORGE CLARK,c WAYMAN E. TURNER,b LUIGI BONSIGNORE,d PAOLO MOCARELLId AND PIER MARIO GERTHOUXd aSchool of Public Health, University of California at Berkeley, Berkeley, CA, USA bDivision of Environmental Health Laboratory Science, National Center for Environmental Health, Centers for Disease Control and Prevention, Atlanta, GA, USA cXenobiotic Detection Systems, Inc., Durham, NC, USA dDepartment of Laboratory Medicine, University of Milano-Biococca, School of Medicine, Hospital of Desio, Desio-Milano, Italy

Polychlorinated dibenzo-p-dioxins (PCDDs), polychlorinated dibenzofurans (PCDFs), and polychlorinated biphenyls (PCBs) are widespread environmental contaminants that exist as complex mixtures and are frequentlydetected at part-per-trillion (ppt) levels in humans. Using isotope dilution high-resolution gas chromatography/ high-resolution mass spectrometry (HRGC/HRMS), we measure the PCDDs, PCDFs, and PCBs in serum of a population of 78 women residing in an area near Seveso, Italywhere a TCDD explosion occurred in 1976 and where furniture is manufactured. The average total dioxin-like toxic equivalents (TEQ) of these women was 25.3 ppt, lipid-adjusted, comparable to other parts of Europe. TCDD levels, however, were higher among the few women who resided in the exposed area in 1976. We examined the possibilityof using the CALUX (chemical- activated luciferase gene expression) bioassayto estimate total TEQ in a small volume of plasma from this population. A total of 32 archived plasma specimens were selected for CALUX bioassay,based on the distribution of Total TEQ byHRGC/HRMS. The CALUX bioassaywas performed blind to HRGC/HRMS results with 2 ml plasma per replicate analysis. Of 32 samples, 10 were below detection limits in the CALUX bioassay. For the 32 samples, the CALUX-TEQ averaged 25.4 ppt, lipid-adjusted (range: 0–127.6) and was not signiﬁcantlydifferent from the HRGC/HRMS Total TEQ average of 31.2 ppt, lipid-adjusted (range: 12.7–88.3) (t ¼ 0.88, P ¼ 0.38), however, the two measures were not signiﬁcantlycorrelated ( Rs ¼ 0.04, P ¼ 0.82). More validation of the CALUX bioassaywith larger sample volume is needed before application as an exposure measure in large-scale epidemiologic studies of health effects of dioxin-like compounds. Journal of Exposure Analysis and Environmental Epidemiology (2005) 15, 310–318. doi:10.1038/sj.jea.7500407 Published online 22 September 2004

Keywords: TCDD, TEQ, CALUX, PCB, PCDD, PCDF.

Introduction toxicity is via binding with the aryl hydrocarbon receptor (AhR) (Whitlock, 1993). Several PCDDs, PCDFs, and PCBs Polychlorinated dibenzo-p-dioxins (PCDDs), dibenzofurans have been shown to cause toxic effects similar to TCDD (PCDFs), and biphenyls (PCBs) constitute a group of (IARC, 1997). Like TCDD, these other dioxins and dioxin- polyhalogenated aromatic hydrocarbons that are persistent, like compounds bind to the AhR with a relative binding widespread environmental contaminants, frequentlydetected capacitythat correlates with toxicityof effects (IARC, 1997). at the parts-per-trillion (ppt) level in lipid tissues of animals Since TCDD and other dioxin-like compounds exist as and humans (Zook and Rappe, 1994). 2,3,7,8-Tetrachlor- complex mixtures of various congeners throughout the odibenzo-p-dioxin (TCDD) is the most toxic congener within environment, calculating total TCDD toxic equivalent this group of compounds and has been associated with a wide (TEQ) concentration is the most relevant exposure measure varietyof health effects including dermal, immunologic, in studies of health effects of dioxins and dioxin-like carcinogenic, reproductive, developmental and endocrinolo- compounds (Van den Berg et al., 1998; Farland et al., gic effects in animals and humans (Birnbaum, 1994, 1995; 2000). In particular, in background-exposed populations, IARC, 1997). It is believed that the mechanism of TCDD these other congeners maycontribute a large portion of TEQ relative to TCDD. Numerous exposure data have been published for residents of countries throughout the world, 1. Address all correspondence to: Dr. Marcella Warner, School of Public including Europe, North America, and Japan (Patterson Health, Universityof California, 2150 Shattuck Avenue, Suite 600, et al., 1994; Papke 1998; Falk et al., 1999; Longnecker et al., Berkeley, CA 94720-7380, USA. Tel.: þ 1-510-642-9544. Fax: þ 1-510- 642-9083. E-mail: [email protected] 2000). These data suggest there are regional differences in Received 8 September 2003; accepted 19 July2004; published online 22 congener proﬁles of TEQ exposure and these differences may September 2004 be due to variations in diet as well as area industrial sources. Dioxin-like TEQ of Seveso women Warn er et al .

The Seveso area of Italy, north of Milan, is a worldwide define the parameters for its use, especiallywith respect to leading furniture-manufacturing area where potential for serum volume needed. exposures exist from treated wood sawdust as well as the The purpose of this studyis two-fold: to characterize location of a historic TCDD explosion in 1976 (Mocarelli current background exposure to PCDDs, PCDFs, and PCBs et al., 1988). Despite these potential sources of TEQ in the Seveso area of Italy, known for furniture manufactur- exposure, the current exposure profile of TEQ congeners ing and the 1976 TCDD explosion based on HRGC/HRMS has not been described in this part of Italy. measurements; and to examine the correlation of these To date, analytical methods to quantify exposure to these measurements with the CALUXs byXDS bioassaymeasure chemicals in biologic samples have included verysensitive of TEQ in this population using a relativelysmall volume of and specific techniques such as isotope-dilution high-resolu- plasma. tion gas chromatography/high-resolution mass spectrometry (HRGC/HRMS). However, these methods require relatively largesamplevolumes(B15 ml). Recently, an AhR-depen- Methods dent recombinant bioassaywas developed that can estimate the total TEQ concentration of a complex mixture in human Study Population serum (Aarts et al., 1996; Denison et al., 1996; Garrison Studyparticipants included women who participated in et al., 1996). The bioassayreportedlyrequires a relatively an endometriosis case-control studyat Desio Hospital, Italy, smaller sample volume (o5 ml). about 25 km north of Milan (Eskenazi et al., 2001). To be The chemical-activated luciferase gene expression (CA- eligible, a woman had to be 20–50 years old, and scheduled LUX) bioassaymeasures the abilityof a chemical mixture to to undergo laparoscopyfor pelvic pain, infertility,tubal activate AhR-dependent gene expression of the firefly ligation, or adnexal/uterine mass at the Hospital of Desio, luciferase gene in geneticallymodified cell lines (Aarts et al., between July1998 and December 1999. As part of the study, 1996; Denison et al., 1996; Garrison et al., 1996). Two women were interviewed bytrained nurse-interviewers and companies perform the CALUX bioassaywith different asked about sociodemographic information, personal habits, cell lines and clean-up procedures. Xenobiotic Detection and reproductive history. Because the Hospital of Desio was Systems (XDS) use a genetically engineered mouse cell the hospital that followed the exposed residents after the 1976 line with a clean-up procedure that specificallyisolates TCDD explosion in Seveso (Mocarelli et al., 1988) and is PCDDs and PCDFs from dioxin-like PCBs and located within 10 km of Seveso, we also asked in what estimates TEQ activityfor each of these classes of dioxin- exposure zone (A, B, R, non-ABR) (Mocarelli et al., 1992) like chemicals (CALUXs byXDS). BioDetection Systems theyhad resided at the time of the explosion. A total of 78 uses a geneticallyengineered rat cell line without separation women provided a 70-ml blood sample. Informed consent of dioxin-like PCBs for estimation of TEQ activity was obtained from participants. (CALUXs byBDS). The CALUXs byBDSbioassayhasbeenusedtoestimate HRGC/HRMS Chemical Analysis serum TEQ in epidemiologic studies of endometriosis Serum samples were aliquoted and stored at À201Cat (Pauwels et al., 2001), immunologic markers (Van Den Hospital of Desio. For each woman, a 15-ml serum sample Heuvel et al., 2002), and sexual maturation (Den Hond et al., was analyzed at US Centers for Disease Control and 2002; Nawrot et al., 2002). However, limited comparison Prevention (CDC) for PCDDs, PCDFs, and PCBs by studies of the CALUXs byBDS bioassaywith the chemical HRGC/HRMS (Patterson et al., 1987). Serum samples were analysis measurements have been completed (Pauwels et al., analyzed for the 17 PCDD/PCDF toxic congeners and 2000; Koppen et al., 2001). Two studies have been completed coplanar PCBs (PCB 77, 81, 126, 169). In addition, 36 PCBs comparing the CALUXs byXDS bioassaywith HRGC/ were measured including the mono-ortho PCBs (PCB 105, HRMS data (Kayama et al., 2001, 2002; Van Wouwe et al., 118, 156, 157, 167). Serum total lipids were calculated using 2003). While the results of these individual studies have an enzymatic ‘summation’ method (Akins et al., 1989). reported significant correlations between CALUX-TEQ and Values were reported on a lipid-weight basis in ppt for chemical data (range of r ¼ 0.43–0.57), each studyhas used PCDDs, PCDFs, and coplanar PCBs, and parts per billion a different volume of serum for the CALUX bioassay (ppb) for the 36 non-coplanar PCBs. To aid in the statistical (ranging from 1 to 10 ml) and variable chemical analysis analysis of data from truncated frequency distributions, methods (GC/ECD, GC/MS, HRGC/HRMS), not always quantifiable results less than the respective method detection measuring all relevant PCDDs, PCDFs, or PCB congeners limits were reported when observed for individual analytes. (Pauwels et al., 2000; Kayama et al., 2001, 2002; Koppen Individual analyte data with non-detectable values were et al., 2001; Van Wouwe et al., 2003). Thus, while there assigned a value equal to one-half the detection limit for is interest in using the CALUX bioassayas a measure of summarystatistics of individual analyteconcentration TEQ in epidemiologic studies, there is still a need to better (Hornung and Reed, 1990).

Journal of Exposure Analysis and Environmental Epidemiology (2005) 15(4) 311 Warner et al. Dioxin-like TEQ of Seveso women

CALUX Bioassay (oPCB-TEQ) and Total TEQ, using the TEF scheme of the WHO (Van den Berg et al., 1998). For calculations of Sample selection Based on the distribution of total TEQ TEQ, all nondetectable data were reassigned a value of zero. data derived from the HRGC/HRMS data for the 78 In addition, REP-TEQs were estimated from CALUXs women, we selected a total of 32 plasma samples (4 ml) to be byXDS Relative Potency(REP) values instead of TEF sent in two shipments to Xenobiotic Detection System, Inc. values (Brown et al., 2001a). CALUXs byXDS REPs are (Durham, NC, USA) for CALUXs byXDS bioassay generallyconsistent with WHO-TEFs. Differences between analysis. Specifically, we randomly selected eight samples per REPS and TEFs appear to be related to the fact that WHO quartile of Total TEQ exposure, ranging from 12.5 ppt TEF values incorporate uptake and metabolism while these (reported detection limit of the CALUX bioassaybased on processes do not have a significant impact on the respon- 4 ml of plasma) to 88.3 ppt. Each shipment included samples siveness of the CALUX bioassay(Brown et al., 2001a). from all four quartiles. XDS was blind to the HRGC/HRMS Statistical analyses were performed using STATA 7.0 results. (Stata, 2001). The distributions of individual PCDD, PCDF, and PCB congener data, as well as TEQ data, were initially Sample preparation Samples were prepared as described examined graphicallyand with standard descriptive statistics. previously(Brown et al., 2002a, 2002b). Briefly,plasma Analysis of variance was used to examine the relation of TEQ samples (4 ml) were processed to remove potential interfering levels with covariates. Because these data were not normally compounds and to separate halogenated dioxin-like PCBs distributed, the relationship between PCDD-TEQ, PCDF- and halogenated dioxins/furans using XCARB affinity TEQ, PCB-TEQ, Total TEQ, and CALUX-TEQ was matrix (Brown et al., 2002a, b). Plasma samples were examined bynonparametric correlation analysis. We con- transferred to hexane-rinsed glass 50 ml centrifuge tubes sidered these relations in three ways. First, we included the with PTFE-lined caps then shaken and extracted four times full sample (n ¼ 32) and assigned a value of zero to all with an acetone/hexane mixture. The pooled extracts were CALUX non-detects. Second, we included the full sample evaporated under nitrogen and the dried residues were (n ¼ 32) and assigned a value of one-half the detection limit further cleaned up bysuspension in hexane and application to all CALUX non-detects. Third, we excluded CALUX to an acid-silica column in series with an XCARB column non-detects from the analysis (n ¼ 22). We repeated the containing activated carbon (Brown et al., 2002a, b). The above analyses for just the DOX fraction of the CALUX acid–silica column removes unwanted organic compounds bioassayresults. We also repeated the above analysesfor each including PAHs. The XCARB column was then differentially CALUX batch separately. Finally, we examined the relation eluted to yield two fractions: a PCB fraction and a PCDD/ of CALUX-TEQ with TEQ estimated from CALUXs by PCDF (DOX) fraction. XDS Relative Potency(REP) values (REP-TEQ) (Brown For the CALUXs byXDS bioassay,a genetically et al., 2001a). engineered cell line is used, which contains the firefly luciferase gene under trans-activational control of the AhR (Garrison et al., 1996). One half of the purified sample Results extracts (B2 ml) in DMSO were suspended in cell culture medium prior to dosing monolayers of H1L6.1 mouse The 78 women in the studywere an average age of 36.3 years hepatoma cells that were grown in 96-well culture plates and (SD ¼ 7.2), ranging from 20 to 49. The average bodymass the analysis was replicated. In addition to the samples, a index was 22.7 mg/kg2 (SD ¼ 3.8). A majorityof the women standard curve of TCDD was generated on each plate. The were parous (n ¼ 44, 60%) and most had a historyof plates were incubated for optimal expression of luciferase lactation (n ¼ 31 of 44, 70%). From the sample of 78 activityin a humidified CO 2 incubator. After incubation, the women, we selected 32 samples for CALUX analysis. medium was removed and the cells were examined micro- Descriptive characteristics of these 32 women were similar scopicallyfor viability.The induced luciferase activitywas to the full sample (mean7SD age: 37.876.0 years, quantified using the luciferase assaykit (Promega). The data mean7SD BMI: 23.173.7 kg/m2, percent parous (n ¼ 22 for the TCDD standard curve were fit to a sigmoid curve of 32, 68%), historyof lactation ( n ¼ 22 of 22, 100%)). described bythe four parameter Hill equation using least- In Table 1, the results of serum HRGC/HRMS analysis squares best-fit modeling. CALUX-TEQ values were calcu- are summarized for all 78 women. The average serum sample lated for plasma samples byinterpolation. weight for the 78 women was 14.4 g serum. TCDD was detected in 19% of the samples with a median of 1.6 ppt, Statistical Analysis lipid-adjusted (range: 0.7–18.9). The four most prevalent Using the HRGC/HRMS data, TEQs were calculated analytes detected included octachlorodibenzodioxin separatelyfor PCDDs (PCDD-TEQ), PCDFs (PCDF- (OCDD), 2,3,4,7,8-pentachlorodibenzofuran (2,3,4,7,8- TEQ), coplanar PCBs (cPCB-TEQ), mono-ortho PCBs PCDF), 3,30,4,40,5-pentachlorobiphenyl (PCB 126), and

312 Journal of Exposure Analysis and Environmental Epidemiology (2005) 15(4) Dioxin-like TEQ of Seveso women Warn er et al .

Ta bl e 1 . Summaryof PCDDs/PCDFs/PCBs byHRGC/HRMS in serum of females ( n ¼ 78), Northern Italy, 1999.

Average detection % Detected Analyte concentration Analyte TEQ ppt limit (lipid-adjusted) (lipid-adjusted) Analyte TEF Median (range) Mean (range)

Dioxins (ppt) 2378-TCDD 1 2.8 19 1.6 (0.7–18.9) 1.6 (0–18.9) 12378-PeCDD 1 2.9 60 3.2 (1.1–9.5) 2.9 (0–9.5) 1234/678-HxCDD 0.1 4.4 88 17.0 (1.6–50.4) 1.7 (0–5.0) 123789-HxCDD 0.1 5.6 6 2.7 (1.1–5.9) 0.2 (0–0.6) 1234678-HpCDD 0.01 44.0 4 13.5 (0.1–79.6) 0.1 (0–0.8) 1234679-HpCDD 0 20.3 0 10.2 (6.5–16.3) 0 (F) OCDD 0.0001 88.4 99 292.0 (34.6–2000) 0.03 (0–0.2)

Furans (ppt) 2378-TCDF 0.1 3.0 1 1.4 (1.0–3.2) 0.009 (0–0.2) 12378-PeCDF 0.05 2.8 1 1.4 (0.3–2.8) 0.003 (0–0.1) 23478-PeCDF 0.5 2.5 95 10.9 (1.4–31.0) 5.7 (0–15.5) 1234/678-HxCDF 0.1 4.3 85 6.0 (2.2–16.6) 0.6 (0–1.7) 123789-HxCDF 0.1 3.1 0 1.3 (0.1–2.9) 0.01 (0–0.06) 234678-HxCDF 0.1 3.0 9 1.5 (0.4–4.9) 0.1 (0–0.5) 1234678-HpCDF 0.01 18.2 1 6.0 (0.1–38.3) 0.02 (0–0.4) 1234789-HpCDF 0.01 3.3 0 1.5 (0.3–4.2) 0.0006 (0–0.03) OCDF 0.0001 FFF(F) F (F)

Coplanar PCBs (ppt) PCB 77 0.0001 FFF(F) F (F) PCB 81 0.0001 14.8 0 7.3 (0.1–11.6) 0 (F) PCB 126 0.1 15.7 91 55.0 (5.0–359.0) 6.4 (0–35.9) PCB 169 0.01 15.3 94 51.4 (6.1–139.0) 0.5 (0–1.4)

Mono-ortho-PCBs (ppb) PCB 105 0.0001 13.2 3 3.8 (0.8–27.8) 0.06 (0–2.8) PCB 118 0.0001 12.1 77 22.0 (2.3–212.0) 2.4 (0–21.2) PCB 156 0.0005 13.2 32 10.0 (0.7–98.7) 3.1 (0–49.4) PCB 157 0.0005 13.1 0 2.4 (0.3–9.1) 0.001 (0–0.004) PCB 167 0.00001 13.1 1 3.8 (0.4–38.2) 0.005 (0–0.4) PCB 189 0.0001 13.1 0 1.7 (0.3–11.8) 0.0002 (0–0.001)

Di-ortho-PCBs (ppb) PCB 138 0 56.6 40 50.5 (0.2–951.0) 0 (F) PCB 153 0 59.9 82 142.0 (8.7–1520.0) 0 (F) PCB 180 0 29.8 90 113.5 (5.4–709.0) 0 (F)

PCDD TEQ 6.6 (0–32.0) PCDF TEQ 6.4 (0–17.1) Coplanar PCB TEQ 6.9 (0–36.7) Mono-ortho-PCBs 5.5 (0–73.7)

PCDD/PCDF TEQ 12.9 (0–49.1) PCDD/PCDF/cPCB TEQ 19.8 (0–64.0) Total TEQ 25.3 (0–88.3)

3,30,4,40,5,50-hexachlorobiphenyl (PCB 169), all of which of similar median concentrations, 55.0 and 51.4 ppt, lipid- were detected in more than 90% of samples. OCDD adjusted, respectively. contributed the highest concentration of the dioxin congeners As presented in Table 1, the total TEQ for the study with a median of 292 ppt, lipid-adjusted. Among furans, the population averaged 25.3 ppt, lipid-adjusted (range: 0–88.3). highest concentration was for 2,3,4,7,8-PCDF (med- In total, the percent contribution to Total TEQ bythe four ian ¼ 10.9 ppt, lipid-adjusted). PCB 126 and PCB 169 were categories of compounds was similar, PCDDs (26.1%),

Journal of Exposure Analysis and Environmental Epidemiology (2005) 15(4) 313 Warner et al. Dioxin-like TEQ of Seveso women

PCDFs (25.3%), coplanar PCBS (27.3%), and mono-ortho higher than the 66 women from the nonexposed zone PCBs (21.7%). The average TEQ values for the four (median ¼ 1.5 ppt, range: o1.5–18.0 ppt) (P ¼ 0.002). Total categories of dioxin-like compounds that contribute to total TEQ levels, however, were not significantlydifferent (Zones TEQ (PCDDs (6.6 ppt), PCDFs (6.4 ppt), coplanar PCBs A, B, R (mean (range) ¼ 34.8 ppt (10.9–63.5)) versus Zone (6.9 ppt) and mono-ortho PCBs (5.5 ppt) were similar but a non-ABR (n ¼ 66, mean (range) ¼ 24.2 ppt (0.3–88.3)), wider range was found for mono-ortho PCBs (0–73.7 ppt). P ¼ 0.11). The PCDD congeners contributing most to TEQ con- In univariate regression, total TEQ was higher among the centrations were 1,2,3,7,8-PCDD (mean ¼ 2.9 ppt TEQ), 78 women who were older (P ¼ 0.01), married (P ¼ 0.01), 2,3,7,8-TCDD (mean ¼ 1.6 ppt TEQ), and 1,2,3,4/6,7,8- parous (P ¼ 0.05), and had a historyof lactation ( P ¼ 0.02). HxCDD (mean ¼ 1.7 ppt TEQ). 2,3,4,7,8-PCDF In multivariate regression, however, onlyage remained a (mean ¼ 5.7 ppt TEQ) contributed the most to PCDF- significant predictor of total TEQ in this population. TEQ. The most dominant coplanar PCB contributing to CALUXs byXDS bioassayresults for the two fractions PCB-TEQ was PCB 126 (mean ¼ 6.4 ppt TEQ), as reported (DOX and PCB) are described for each of the two shipments byothers (Kiviranta et al., 2002). It is noteworthythat two separately. For the first shipment of 16 samples, with 2 ml of congeners of mono-ortho PCBs (PCB 118, PCB 156) sample extract, CALUX-TEQ for the DOX fraction was individuallycontributed more to total TEQ (mean ¼ 2.4 ppt determined for three of the 16 samples. The remaining 13 of and mean ¼ 3.1 ppt, respectively) than TCDD. 16 samples were below the limit of quantification (LOQ) s The relative percent contribution to Total TEQ by (NDo17 ppt, lipid-adjusted) in the CALUX byXDS congeners of PCDDs, PCDFs, and PCBs for the group is bioassay. For six of 13 nondetectable samples, however, the presented in Figure 1. TCDD contributes 4.5% to the total results were close enough to the LOQ level to estimate the TEQ of this population. The major congeners that contribute CALUX-TEQ. For the remaining seven nondetectable to TEQ in this population are 1,2,3,7,8-PCDD (11.5%), samples, the average detection limit was 14.8 ppt, lipid- 2,3,4,7,8-PCDF (25.1%), and PCB 126 (26.6%). Other adjusted. For the analysis of the PCB fraction, all 16 samples notable congeners that contribute more than 5% to the total were nondetectable with an average detection limit of TEQ include 1,2,3,4/6,7,8-HxCDD (7.3%), 2,30,4,40,5- 20.4 ppt, lipid-adjusted. Replication of the analysis with the PeCB (PCB 118) (9.5%), and 2,3,30,4,40,5-HxCB (PCB remaining 2 ml of sample extract was not possible due to 156) (7.5%). bacterial contamination of the analytical plate. We had information for 73 of the 78 women in this study For the second shipment of 16 samples, replication of the on zone of residence at the time of the 1976 TCDD explosion analysis with remaining sample extract was possible and in Seveso. The majorityof women resided in the nonexposed results meeting optimal QA/QC criteria were reported. With area, Zone non-ABR (n ¼ 66, 90%), but seven women 2 ml per sample, the CALUX-TEQ for the DOX fraction reported theylived in one of the exposed areas (Zone A was determined for 13 of the 16 samples. The remaining (n ¼ 1); Zone B (n ¼ 2), Zone R (n ¼ 4)). The TCDD levels three samples were nondetects in the CALUXs byXDS of the seven women from the exposed areas (med- bioassaywith an average detection limit of 12.3 ppt, lipid- ian ¼ 4.3 ppt, range ¼ o2.3–18.9 ppt) were significantly adjusted. For the analysis of the PCB fraction, 14 of 16

30%

25%

20%

15%

10%

D D D D D F F F F F F 7 1 8 9D 8 8 9F 8F 26 69 05 18 56 57 67 89 78 78 8 78 79 78 78 8 7 78 89 7 8 3 37 6 6 3 37 47 6 7 B B 1 1 1 1 1 1 1 1 2 2 /6 37 OCDD 2 2 3 /6 37 46 OCDF C C B B B B B B B B 1 34 2 34 34 1 2 34 2 3 34 34 P P C C C C C C C C 2 1 2 2 2 1 2 2 2 P P P P P P P P 1 1 1 1 1 1 Figure 1. Relative percent contribution to TEQ byPCDDs/PCDFs/PCBs ( n ¼ 78).

314 Journal of Exposure Analysis and Environmental Epidemiology (2005) 15(4) Dioxin-like TEQ of Seveso women Warn er et al . samples were nondetectable with an average detection limit of the data assigning CALUX nondetects a value of one-half 33.0 ppt, lipid-adjusted. the detection limit, but the results were similar (data We considered each shipment separatelyand together in not shown). the analysis and although the second shipment had fewer Given that the CALUX PCB fraction was detected in only nondetects, the results were not different (data not shown). two of 32 samples, we also examined the relation of just the Because the results are similar for both shipments, the CALUX DOX fraction with the HRGC/HRMS data. For CALUX-TEQ results are summarized together. In Table 2 the full sample, the CALUX DOX fraction TEQ averaged we present results in two ways: (1) including the full sample 21.2 ppt (range: 0–67.3) compared with the HRGC/HRMS- (n ¼ 32) but assigning CALUX (DOX fraction) nondetects a derived PCDD/PCDF TEQ average of 14.7 ppt (range: 5.2– value of zero, and (2) excluding CALUX (DOX fraction) 49.1) (t ¼À1.8, P ¼ 0.07). Excluding nondetects (n ¼ 22), nondetects (n ¼ 22). For the full sample (n ¼ 32), the the CALUX DOX fraction TEQ averaged 30.8 ppt (range: CALUX-TEQ averaged 25.4 ppt, lipid-adjusted (range: 0– 1.6–67.3) and was significantlydifferent from the HRGC/ 127.6), not significantlydifferent from the HRGC/HRMS- HRMS-derived PCDD/PCDF TEQ average of 15.6 ppt derived total TEQ average of 31.2 ppt (range: 12.7–88.3) for (range: 5.2–49.1) (t ¼À4.0, Po0.001). In Figure 2, we the same 32 samples (t ¼ 0.88, P ¼ 0.38). Excluding non- graphicallypresent the relationship of the CALUX DOX detects (n ¼ 22), the CALUX-TEQ averaged 37.0 ppt, lipid- fraction TEQ versus HRGC/HRMS-derived PCDD/PCDF adjusted (range: 1.6–127.6) and were not significantly TEQ (Rs ¼ 0.31, P ¼ 0.08). The results of correlation different from the HRGC/HRMS-derived total TEQ of analysis of the CALUX DOX fraction TEQ versus 31.9 ppt (range: 12.7–88.3) for the same 22 samples HRGC/HRMS data did not differ from the CALUX-TEQ (t ¼À0.62, P ¼ 0.54). For the 10 excluded non-detects, the results (see Table 2). HRGC/HRMS-derived total TEQ averaged 29.6 ppt, lipid- We also considered the relation of the CALUX DOX adjusted (range 16.6–64.0). fraction TEQ with TEQ estimated from CALUXs byXDS As presented in Table 2, for the full sample of 32, the REP values instead of TEF values (REP-TEQ). REP values nonparametric correlation (Rs) of CALUX-TEQ with differ slightlyfrom TEF values primarilybecause uptake and HRGC/HRMS total TEQ was 0.04 (P ¼ 0.82). The highest tissue distribution are not relevant to the response of the correlations, although not significant, were found for PCDD compounds for a cell-based bioassay(Brown et al., 2001a).

TEQ (Rs ¼ 0.34, P ¼ 0.06), PCDF TEQ (Rs ¼ 0.25, P ¼ The total TEQ was reduced somewhat byapplying the REP 0.17) and PCDD/PCDF TEQ (Rs ¼ 0.31, P ¼ 0.09). When values to the HRGC/HRMS data. The Total REP-TEQ nondetects were excluded, the correlations did not change averaged 18.2 ppt (range: 7.4–55.0) for the full sample

(PCDD Rs ¼ 0.33, P ¼ 0.13; PCDF Rs ¼ 0.15, P ¼ 0.50; (n ¼ 32), and 18.9 ppt (range: 7.4–55.0) excluding the PCDD/PCDF Rs ¼ 0.24, P ¼ 0.28). We also examined nondetects (n ¼ 22). Overall the results of correlation analysis

Ta bl e 2 . Results of nonparametric correlation analysis between HRGC/HRMS- and CALUX-derived TEQ data.

Full sample (n ¼ 32) Excluding non-detects (n ¼ 22)

TEQ CALUX total CALUX DOX fraction TEQ CALUX total CALUX DOX TEQ versus TEQ versus TEQ versus fraction TEQ HRGC/HRMS HRGC/HRMS HRGC/HRMS versus HRGC/HRMS

Mean (range) Rs (P-value) Rs (P-value) Mean (range) Rs (P-value) Rs (P-value)

CALUX TEQa,b Total TEQ 25.4 (0–127.6) 37.0 (1.6–127.6) DOX fraction TEQ 21.2 (0–67.3) 30.8 (1.6–67.3) HRGC/HRMS-Derived-TEQ PCDD TEQ 7.6 (0.1–32.0) 0.34 (0.06) 0.34 (0.06) 8.2 (0.1–32.0) 0.33 (0.13) 0.34 (0.12) PCDF TEQ 7.1 (3.6–17.1) 0.25 (0.17) 0.25 (0.17) 7.4 (3.6–17.1) 0.15 (0.50) 0.16 (0.48) cPCB TEQ 8.1 (2.6–36.7) À0.18 (0.34) À0.16 (0.37) 7.3 (2.6–14.9) À0.17 (0.45) À0.15 (0.52) ortho PCB TEQ 8.3 (0–73.7) À0.21 (0.25) À0.20 (0.28) 9.0 (0–73.7) À0.25 (0.25) À0.23 (0.30)

PCDD/PCDF TEQ 14.7 (5.2–49.1) 0.31 (0.09) 0.31 (0.08) 15.6 (5.2–49.1) 0.24 (0.28) 0.24 (0.28) PCDD/PCDF/cPCB TEQ 22.8 (9.1–64.0) 0.07 (0.70) 0.08 (0.67) 22.9 (9.1–64.0) 0.04 (0.88) 0.05 (0.82) Total TEQ 31.2 (12.7–88.3) 0.04 (0.82) 0.05 (0.78) 31.9 (12.7–88.3) 0.001 (0.99) 0.02 (0.93) aCALUX DOX fraction: % detected ¼ 22/32 ¼ 69%, average DL ¼ 15.2 ppt, lipid-adjusted. bCALUX PCB fraction: % detected ¼ 2/32 ¼ 6%, average DL ¼ 26.7 ppt, lipid-adjusted.

Journal of Exposure Analysis and Environmental Epidemiology (2005) 15(4) 315 Warner et al. Dioxin-like TEQ of Seveso women

70 more to total TEQ than TCDD. Finally, consistent with that reported byothers (Svensson et al., 1991; Patterson et al., 60 1994; Papke 1998; Kiviranta et al., 2002), we found that age 50 was a signiﬁcant predictor of Total TEQ in this Italian

40 sample. In this study, the serum TCDD levels of the seven women 30 Batch 1 who resided in exposed areas at the time of the 1976 20 Batch 2 explosion in Seveso were signiﬁcantlyhigher than the women from the nonexposed area. This ﬁnding is consistent with the 10 onlyother studyof contemporaryTCDD levels of Seveso CALUX DOX TEQ (ppt, lipid-adjusted) CALUX DOX TEQ (ppt, 0 residents (Landi et al., 1998). Landi et al. (1998) measured 0 10 20 30 40 50 TCDD levels in plasma collected in 1992–1994, from 30

HRGC/HRMSPCDD/PCDF TEQ (ppt, lipid-adjusted) women from exposed areas (Zones A and B) and 32 women from the nonexposed area (Zone non-ABR). The TCDD Figure 2. Scatterplot of CALUX DOX fraction TEQ versus PCDD/ levels of the women in our studywho resided in the non- PCDF TEQ byBatch ( n ¼ 32, Rs ¼ 0.31, P ¼ 0.08). exposed area (Zone non-ABR) in 1976 are comparable to those reported byLandi et al. (1998). The TCDD levels of were similar (data not shown). The nonparametric correla- the women in 1998-1999 from the exposed areas, Zones A, B tion of the CALUX DOX fraction TEQ with total REP- and R, (median ¼ 4.3 ppt), are lower than those reported by TEQ was 0.21 (P ¼ 0.24) and with PCDD/PCDF REP-TEQ Landi et al. (median ¼ 16.8 ppt) in 1992–1994. However, was 0.31 (P ¼ 0.08). four of the women in our studyresided in Zone R, a zone We calculated the sensitivityand specificityof the CALUX that had less exposure than Zones A and B. bioassayto correctlycategorize HRGC/HRMS-derived total PCDDs, PCDFs, and PCBs exist as complex mixtures and TEQ for a wide range of exposure cut-points (15, 20, 35, 40, have been shown to cause toxic effects similar to TCDD, thus 50, 60 ppt). Sensitivity(range: 20–54%) and specificity it is important to include them in exposure assessment in (range: 50–93%) both varied across the range of exposure studies of health effects of dioxin-like compounds. We cut-points considered. The 15-ppt cut-point had the highest examined the potential of using the CALUX bioassayto sensitivity(54%), but specificitywas 50%. estimate TEQ in a small volume of plasma. We found no As an additional test, three blind QA/QC 5-ml serum significant correlation between CALUX-TEQ and total TEQ samples (NIST SRM 1589) were analyzed by the CALUXs or anyTEQ measure derived from HRGC/HRMS data in byXDS bioassay(NIST, 2000). The total TEQ for the this sample of Italian women with background exposure to samples based on HRGC/HRMS averaged 34.2 ppt, lipid- dioxin and dioxin-like compounds. If more plasma were adjusted (SD ¼ 3.4). The CALUX total TEQ reported for analyzed, the results mayhave improved. In a recent studyin these three samples averaged 78.3 ppt (16.6, 91.7, and Belgium, using 10 ml serum for the CALUXs byXDS 126.7 ppt). bioassay, a significant correlation between HRGC/MS- derived PCDD/PCDF-TEQ and CALUX-TEQ (dioxin fraction only) of 0.71 was reported (Van Wouwe et al., Discussion 2003). We found the use of a limited amount of plasma for the CALUXs byXDS bioassayresulted in a wide variation PCDD, PCDF and PCB concentrations measured in this in the measure. studyof women from the Seveso area of Italywere found to Previous studies using the CALUXs byBDS bioassayall be comparable to bodyburdens reported in other back- report using 1–4 ml serum (Pauwels et al., 2000; Koppen ground-exposed populations in industrial nations (Gonzalez et al., 2001, 2002). However, in this study, using 2 ml of et al., 1998; Papke 1998; Longnecker et al., 2000). Also, the plasma per replicate analysis for the CALUXs byXDS distribution of congeners was similar to that reported for bioassay, 31% of samples were not detectable for the other North American and European countries (Patterson PCDD/PCDF fraction and 94% of the samples were not et al., 1994; Papke 1998) with the four most prevalent detectable for the PCB fraction. Previous studies do not analytes being OCDD, 2,3,4,7,8-PCDF, PCB 126, and PCB describe using as extensive a clean-up procedure as used 169. We also found the four categories of compounds in this study(Brown et al., 2002b); thus, it is possible (PCDD, PCDFs, coplanar PCBs, mono-ortho-PCBs) con- that previous studies maybe including in the measure- tributed a roughlysimilar percent to Total TEQ ( B25%). ment other nondioxin-like compounds that can bind to and TCDD contributed only4.5% to the total TEQ of this activate the AhR (Schecter et al., 1999; Denison, 2000). population. It is noteworthythat two congeners of mono- In fact, Pauwels et al. reported a significant correlation ortho PCBs (PCB 118, PCB 156) individuallycontributed between CALUX-TEQ and the sum of four major PCB

316 Journal of Exposure Analysis and Environmental Epidemiology (2005) 15(4) Dioxin-like TEQ of Seveso women Warn er et al . congeners (118, 138, 153, 180) that are non-dioxin-like (only HRMS method used to measure these compounds in 15 ml of PCB 118 has a TEF value of 0.0001) (r ¼ 0.43, Po0.001). serum, did not correlate well with the CALUX bioassay In this study, we found no evidence of a correlation between using 2 ml of plasma in this population. This ﬁnding points to CALUX-TEQ and sum of the same four PCB congeners the need for further validation before application of CALUX

(Rs ¼ 0.12, P ¼ 0.52). bioassayin large-scale epidemiologic studies. These results The CALUX bioassaycan respond to a wide range of suggest that the amount of serum required for measuring different compounds. Unlike chemical analysis methods, background TEQ levels bythe CALUX bioassaymaybe which measure the concentrations of specific compounds, the larger than previouslythought and sample clean-up mayalso CALUX bioassaysums the effects of all compounds present need to be a prerequisite to its application. in the sample extract that can bind to and activate the AhR. For a crude extract, this mayinclude ‘‘classical’’ AhR ligands such as PCDDs, PCDFs, PCBs, polybrominated dibenzo-p- Acknowledgements dioxins, polybrominated dibenzofurans, polybrominated biphenyls, polybrominated diphenyl ethers, and some poly- We gratefullyacknowledge Stefania Casalini, PhD for aromatic hydrocarbons, as well as ‘‘nonclassical’’ AhR coordinating data collection at Hospital of Desio. We would ligands including bilirubin, tryptophan metabolites, and especiallylike to thank the women who participated in this some corticosteroids (Denison and Heath-Pagliuso, 1998; study.This studywas supported byGrant Numbers R01 Van Wouwe et al., 2003). Because the CALUX bioassay ES07171 and F06 TW02075-01 from the National Institutes cannot discriminate between these different active com- of Health, R82471 from the US Environmental Protection pounds, it is necessaryto include sample clean-up procedures Agency, EA-M1977 from the Endometriosis Association, that can isolate the compounds of interest. Further, high 2P30-ESO01896-17 from the National Institute of Environ- levels of some PCBs have been shown to antagonize the mental Health Sciences, and #2896 from Regione Lombar- dioxins if not separated (Garrison et al., 1996). It has been dia and Fondazione Lombardia Ambiente, Milan, Italy. reported that the TEQ value especiallyin samples containing relativelylarge amount of PCB 118 might be underestimated in the CALUX bioassaycompared to an HRGC/HRMS References determination (Bovee et al., 1998; Laier et al., 2003). Inadequate sample clean-up prior to the CALUX bioassay Aarts J., Cenjin P., and Blankvoort B., et al. Application of the chemical- activated luciferase expression (CALUX) bioassayfor quantification of dioxin- mightresultinthemeasurementofactivityduetothese like compounds in small samples of human milk and blood plasma. interfering compounds, and an inaccurate TEQ report for the Organohalogen Compds 1996: 27: 285–290. sample (Brown et al., 2002b). It is not clear as to what Akins J., Waldrep K., and Bernett J. The estimation of total serum lipids bya completelyenzymatic summation method. Clin Chim Acta 1989: 184: 219– sample clean-up methods have been used in previous studies 226. where higher correlations have been reported (Pauwels et al., Birnbaum L. The mechanism of dioxin toxicity: relationship to risk assessment. 2000; Koppen et al., 2001, 2002). Environ Health Perspect 1994: 102: 157–167. Birnbaum L. Developmental effects of dioxins and related endocrine disrupting Because it has been thought that the CALUX bioassay chemicals. Toxicol Lett 1995: 82/83: 743–750. could estimate exposure using smaller sample volumes and at Bovee T.F., Hoogenboom L.A., Hamers A.R., Traag W.A., Zuidema T., Aarts lower cost, the CALUX bioassayhas also been suggested as J.M., Brouwer A., and Kuiper H.A. Validation and use of the CALUX- bioassayfor the determination of dioxins and PCBs in bovine milk. Food Addit a screening assayto identifythose samples that should be Contamin 1998: 15: 863–875. further analyzed by chemical analysis methods (Brown et al., Brown D., Chu M., Van Overmeire I., Chu A., and Clark G. Determination of 2001a). In order for a screening assayto be useful, it should REP values for the CALUX bioassayand comparison to the WHO TEF values. Organohalogen Compds 2001a: 53: 211–214. have high sensitivityat the expense of specificity.We found Brown D., Goeyens L., Van Overmeire I., Chu M., Murata H., and Clark G. onlymoderate sensitivity(20–54%). Also, we have found Qualitycontrol criteria implemented for monitoring the use of the CALUX that larger sample volumes maybe necessary.The CALUX bioassay. Organohalogen Compds 2001b: 54: 32–35. Brown D., Nakamura M., Chu M., Denison M., Murata H., and Clark G. bioassaymaybe more appropriate as a screening tool for Recoverydeterminations for bioassayanalysis: considerations and results. environmental samples that have higher levels of exposure Organohalogen Compds 2002a: 58: 357–360. than human serum, unless a larger volume of serum is used Brown D., Van Overmeire I., Goeyens L., Chu M., Denison M., and Clark G. Elimination of interfering compounds in preparation for analysis by an Ah (Van Overmeire et al., 2000, 2001; Brown et al., 2001b; receptor based bioassay. Organohalogen Compds 2002b: 58: 401–404. USEPA, 2002). Den Hond E., Roels H.A., and Hoppenbrouwers K., et al. Sexual maturation in In conclusion, we found using HRGC/HRMS that a relation to polychlorinated aromatic hydrocarbons: Sharpe and Skakkebaek’s hypothesis revisited. Environ Health Perspect 2002: 110: 771–776. population of 78 women from the Seveso area of Italyhad an Denison M. Optimization of the CALUX Bioassayfor use in detecting dioxin and average TEQ of 25.3 ppt, lipid-adjusted, comparable to other related chemicals in serum. Superfund Basic Research Program Brief, Number parts of Europe, and that the levels of TCDD were higher in 67, 2000. Denison M.S., and Heath-Pagliuso S. The Ah receptor: a regulator of the the few women who resided in the exposed area at the time of biochemical and toxicological actions of structurallydiverse chemicals. Bull the 1976 TCDD explosion. We found that the HRGC/ EnvironContaminToxicol1998: 61: 557–568.

Journal of Exposure Analysis and Environmental Epidemiology (2005) 15(4) 317 Warner et al. Dioxin-like TEQ of Seveso women

Denison M.S., Rogers W., and Fair M., et al. Application of the CALUX Nawrot T.S., Staessen J.A., and Den Hond E.M., et al. Host and environmental bioassaysystem for the detection of dioxin-like chemicals (Ah receptor ligands) determinants of polychlorinated aromatic hydrocarbons in serum of adoles- in whole serum samples and in extracts from commercial and consumer cents. Environ Health Perspect 2002: 110: 583–589. products. Organohalogen Compds 1996: 27: 280–284. NIST. Certificate of Analysis Standard Reference Material 1589a. National Eskenazi B., Warner M., Bonsignore L., Olive D., Samuels S., and Vercellini P. Institute of Standards and Technology, Gaithersburg, MD, 2000. Validation studyof nonsurgical diagnosis of endometriosis. Fertil Steril 2001: Papke O. PCDD/PCDF: human background data for Germany, a 10-year 76: 929–935. experience. Environ Health Perspect 1998: 106(Suppl 2): 723–731. Falk C., Hanrahan L., and Anderson H.A., et al. Bodyburden levels of dioxin, Patterson D., Hampton L., and Lapeza C., et al. High-resolution gas furans, and PCBs among frequent consumers of Great Lakes sport fish. The chromatographic/high-resolution mass spectrometric analysis of human serum Great Lakes Consortium. Environ Res 1999: 80: S19–S25. on a whole-weight and lipid basis for 2,3,7,8-tetrachlorodibenzo-p-dioxin. Farland W.H., Schaum J., and Winters D., et al. USEPA’s risk characterization Anal Chem 1987: 59: 2000–2005. of dioxin and related compounds. Organohalogen Compds 2000: 48: 248–251. Patterson D.G., Todd G.D., and Turner W.E., et al. Levels of non-ortho- Garrison P.M., Tullis K., Aarts J., Brouwer A., GiesyJ.P., and Denison M.S. substituted (coplanar), mono- and di-ortho-substituted polychlorinated Species-specific recombinant cell lines as bioassaysystems for the detection of biphenyls, dibenzo-p-dioxins, and dibenzofurans in human serum and adipose 2,3,7,8-tetrachlorodibenzo-p-dioxin-like chemicals. Fundam Appl Toxicol tissue. Environ Health Perspect 1994: 102(Suppl 1): 195–204. 1996: 30: 194–203. Pauwels A., Cenijn P., Schepens P., and Brouwer A. Comparison of chemical- Gonzalez C.A., Kogevinas M., and Huici A., et al. Blood levels of activated luciferase gene expression bioassayand gas chromatographyfor PCB polychlorinated dibenzodioxins, polychlorinated dibenzofurans and poly- determination in human serum and follicular fluid. Environ Health Perspect chlorinated biphenyls in the general population of a Spanish Mediterranean 2000: 108: 553–557. city. Chemosphere 1998: 36(3): 419–426. Pauwels A., Schepens P.J.C., and Hooghe T.D., et al. The risk of endometriosis Hornung R., and Reed L. Estimation of average concentration in the presence of and exposure to dioxins and polychlorinated biphenyls: a case–control study of non-detectable values. Appl Occup Environ Hyg 1990: 5: 48–51. infertile women. Hum Reprod 2001: 16: 2050–2055. IARC. Polychlorinated Dibenzo-para-Dioxins and Polychlorinated Dibenzofurans, Schecter A., Sheu S., Birnbaum L.S., DeVito M.J., Denison M.S., and Papke Vol. 69. International Agencyfor Research on Cancer. Lyon, France, 1997. O. A comparison and discussion of two differing methods of measuring dioxin- Kayama F., Hamamatsu A., Sagisaka K., Brown D., Clark G., and Suzuki T. like compounds: gas chromatography-mass spectrometry and the calux CALUX assayis applicable to human epidemiological studies. Organohalogens bioassay- implications for health studies. Organohalogens Compds 1999: 40: Compds 2001: 54: 48–50. 247–253. Kayama F., Horiguchi H., and Hoguma E., et al. Regional differences of blood Stata C Stata Statistical Software: Release 7.0. Stata Press, College Station, dioxin and organochlorine pesticides concentrations of Japanese female TX, 2001. farmers F application of Calux assayfor epidemiological study. Organohalo- Svensson B.G., Nilsson A., Hansson M., Rappe C., Akesson B., and Skerfving gen Compds 2002: 55: 275–278. S. Exposure to dioxins and dibenzofurans through the consumption of fish. N Kiviranta H., Vartiainen T., and Tuomisto J. Polychlorinated dibenzo-p-dioxins, Engl J Med 1991: 324: 8–12. dibenzofurans, and biphenyls in fishermen in Finland. Environ Health Perspect USEPA. Report on the comparison of cell-based assays and mass spectrometry 2002: 110: 355–361. methods for the analysis of PCDDs, PCDFs, and PCBs in biosolids, pp. 1–20. Koppen G., Covaci A., and Van Cleuvenbergen R., et al. Comparison of US Environmental Protection Agency, Office of Water, Office of Science and CALUX-TEQ values with PCB and PCDD/F measurements in human serum Technology, Engineering and Analysis Division. Washington, DC, 2002. of the Flanders Environmental and Health Study(FLEHS). Toxicol Lett Van den Berg M., Birnbaum L., and Bosveld A.T.C.. et al. Toxic equivalency 2001: 123: 59–67. factors (TEFs) for PCBs, PCDDs, PCDFs for humans and wildlife. Environ Koppen G., Covaci A., and Van Cleuvenbergen R., et al. Persistent Health Perspect 1998: 106: 775–792. organochlorine pollutants in human serum of 50–65 years old women in the Van Den Heuvel R.L., Koppen G., and Staessen J.A., et al. Immunologic Flanders Environmental and Health Study(FLEHS). Part 1: Concentrations biomarkers in relation to exposure markers of PCBs and dioxins in Flemish and regional differences. Chemosphere 2002: 48: 811–825. adolescents (Belgium). Environ Health Perspect 2002: 110: 595–600. Laier P., Cederberg T., Larsen J.C., and Vinggaard A.M. Applicabilityof the Van Overmeire I., Clark G., and Brown D., et al. Trace contamination CALUX bioassayfor screening of dioxin levels in human milk samples. Food with dioxin-like chemicals: evaluation of bioassay-based TEQ determination Addit Contamin 2003: 20: 583–595. for hazard assessment and regulatoryresponses. Environ Sci Policy 2001: 4: Landi M.T., Consonni D., and Patterson Jr. D.G., et al. 2,3,7,8- 345–357. Tetrachlorodibenzo-p-dioxin plasma levels in Seveso 20 years after the Van Overmeire I., Goeyens L., and Beernaert H., et al. A comparative study of accident. Environ Health Perspect 1998: 106: 273–277. GC-HRMS and CALUX TEQ determinations in food samples bythe Belgiam Longnecker M.P., Ryan J.J., Gladen B.C., and Schecter A.J. Correlations federal ministries of public health and agriculture. Organohalogens Compds among human plasma levels of dioxin-like compounds and polychlorinated 2000: 45: 196–199. biphenyls (PCBs) and implications for epidemiologic studies. Arch Environ Van Wouwe N., Eppe G., and Xhrouet C., et al. Analysis of PCDD/Fs in human Health 2000: 55: 195–200. blood plasma using CALUX bioassayand GC-HRMS: A comparison. Mocarelli P., Marocchi A., and Brambilla P., et al. Human data derived from Organohalogen Compds 2003: 60: 211–214. the Seveso accident-relevance for human risk assessment. Toxic Subst J 1992: Whitlock J. Mechanistic aspects of dioxin action. Chem Res Toxicol 1993: 6: 12: 51–73. 754–763. Mocarelli P., Pocchiari F., and Nelson N. Preliminaryreport: 2,3,7,8- Zook D, and Rappe C Environmental sources, distribution, and fate. In: tetrachlorodibenzo-p-dioxin exposure to humans F Seveso, Italy. MMWR Schecter A. (Ed.). Dioxins and Health. Plenum Press, New York, 1994, pp. 1988, 733–736. 79–113.

318 Journal of Exposure Analysis and Environmental Epidemiology (2005) 15(4)