Interlaboratory Comparison on POPs in Food 2019. The twentieth round of an international study
Abstract
Summary
The 20th round of the Interlaboratory Comparison on POPs in Food was conducted in 2019 by the Norwegian Institute of Public Health (NIPH). The study included the determination of the 2,3,7,8-chlorinated dibenzo-p-dioxins (PCDDs) and dibenzofurans (PCDFs), as well as dioxin-like non-ortho and mono-ortho chlorinated biphenyls (PCBs) in three different food items. Additionally, the participating laboratories could determine the concentrations of six indicator PCBs, polybrominated diphenyl ethers (PBDEs) and hexabromocyclododecanes (HBCDs) in the same food samples.
For the first time a fourth matrix was added, designated for the determination of the following poly- and perfluoroalkyl substances (PFASs): Perfluorooctanesulfonate (PFOS), perfluorohexanesulfonate (PFHxS), perfluorooctanoate (PFOA), perfluorononanoate (PFNA), perfluorodecanoate (PFDA) and perfluoroundecanoate (PFUnDA).
The objectives of this interlaboratory comparison study were
A. To offer a tool for quality assurance to the participating laboratories
B. To assess the between laboratory reproducibility
C. To assess the readiness of expert laboratories worldwide to determine levels of chlorinated and brominated persistent organic pollutants, as well as for PFASs, in regular foodstuffs.
With respect to PCDDs/PCDFs, non-ortho PCBs, mono-ortho PCBs, indicator PCBs, PBDEs and α-HBCD, the 2019-round of the study was performed on unfortified homogenates of veal, herring and brown meat from crab (hereafter referred to as Brown meat). An unfortified homogenate of perch was offered for the determination of PFASs.
If desired, the laboratories could also determine the concentrations of PCDDs/PCDFs, nonortho PCBs, mono-ortho PCBs, indicator PCBs, PBDEs and α-HBCD in standard solutions from Cambridge Isotope Laboratories, provided by NIPH.
The test materials were sent to 73 laboratories worldwide in January, and results were returned from 72 of these.
A draft report was made available on our webpage www.fhi.no/ILC in July 2019, and the deadline for commenting on the published results was set to September 15th 2019.
This report presents the reported results for all seventeen 2,3,7,8-substituted PCDDs/PCDFs, the four non-ortho substituted PCBs #77, 81, 126 and 169 and the eight mono-ortho substituted PCBs #105, 114, 118, 123, 156, 157, 167, 189 in the three food items on a fresh weight and lipid weight basis.
The results of eight PBDEs #28, 47, 99, 100, 153, 154, 183 and 209, six indicator PCBs #28, 52, 101, 138, 153 and 180, and total HBCDs as well as the α-, β- and γ-isomers are also presented.
For the first time, we will also present the results of PFOS, PFHxS, PFOA, PFNA, PFDA and PFUnDA in a sample of Perch.
The consensus concentration (assigned value) for each analyte in the four food samples was determined as follows: For the seventeen 2,3,7,8-substituted PCDDs/PCDFs, the four nonortho substituted PCBs and the eight mono-ortho substituted PCBs non-detected congeners were assigned a concentration corresponding to the reported detection limits. The median of all reported concentrations for each analyte was then calculated. All values above twice the median were removed from the calculation. The consensus median and consensus mean as well as standard deviation (SD) were calculated from the remaining data, and this second median was called consensus value.
For the PBDEs, the indicator PCBs and HBCDs the non-detects were removed from the data set. The median of all reported concentrations for each analyte was then calculated. All values above twice the median were removed from the calculation. The consensus median (consensus value) and consensus mean as well as standard deviation (SD) were calculated from the remaining data.
When calculating the results for PFASs in the sample of Perch, we adopted the same approach as we used for the indicator PCBs and PBDEs. However, as the number of reported results were much lower for PFAS than for the other groups of analytes, outliers would contribute to a larger degree. Therefor we have chosen to remove low outliers, as well as high outliers, from the PFAS data set.
Toxic equivalents (TEQs) were calculated from the consensus medians of individual congeners using the toxic equivalency factors derived by WHO 2006 (from 2012: WHO2006TEQs as opposed to WHO1998TEQs as in the reports published before 2012).
Z-scores for the PCDD/PCDF TEQs were calculated for each laboratory using ±20 % of the consensus TEQs as a value for target standard deviation (σ=0.2), on both fresh weight and lipid weight basis. Further, Z-scores were calculated for the non-ortho PCB TEQ, the monoortho PCB TEQ, the total TEQ, the sum of six indicator PCBs, the sum of eight PBDEs, total HBCD, and the three isomers of HBCD and for each single congener in all three matrices of Veal, Herring and Brown meat (σ=0.2) (both on a fresh weight and a lipid weight basis). The z-scores for PFASs were calculated in the same way, but on fresh weight base only.
The consensus values of the standard solutions were calculated as mentioned above with the exception of the removal of all values exceeding ±50 % of the median prior to the final calculation of the consensus median and mean.
The consensus values for the lipid contents were calculated by first excluding results deviating more than two SD from the mean of all values, and then re-calculating the median, mean and SD.
The sample of Brown meat was the sample in this study with the highest total TEQ (2.3 pg TE/g fw, WHO2006TEFs). For this sample, Z-scores within ±1 were obtained by 82 % of the participating laboratories, and Z-scores within ±2 were achieved by 93 % of the participants (Z scores within ±1 and ±2, corresponding to a trueness of ±20 % and ±40 %, respectively). The Brown meat -sample was closely followed by the Herring-sample, with a consensus total TEQ of 2.1 pg TE/g fw (WHO2006TEFs). For this sample 86 % of the participants achieved Z-scores within ±1, and an impressing 98 % achieved Z-scores within ±2!
For the sample of Veal, which on beforehand was assumed to be the sample with lowest level of contamination (total TEQ=0.082 pg TE/g, WHO2006TEFs, fresh weight), Z-scores within ±1 were obtained by 33 % of the reporting participants on fresh weight basis, and Z-scores within ±2 was achieved by 53 % of the participants. This clearly mirrors the high number of non-detects reported for the PCDD/PCDFs, and the variation in the reported limits of detection reported by the participants.
The relative standard deviation (RSD) calculated for the total TEQ after removal of outliers was 15-23 % for the two samples that had the highest levels of contamination (Herring and Brown meat , respectively). For the lower contaminated Veal-sample, the corresponding RSD was 115 %. Considering the levels of contamination in the three food samples in this study, it may be concluded that the abilities and performance of laboratories worldwide in determining dioxin-like compounds is generally good.
For the food samples, 43-56 laboratories reported results for the six indicator PCBs. The total consensus concentrations on fresh weight basis for six indicator PCBs were 300 pg/g (10 %) in Veal, 12201 pg/g (16 %) in Herring and 5837 pg/g (13 %) in Brown meat, with total RSDs given in parentheses.
16-27 laboratories reported concentrations for all seven of the tetra- to hepta-PBDEs and 14-21 laboratories reported concentrations for PBDE-209.
The consensus concentrations of the sum of the PBDEs with PBDE-209 excluded, were on fresh weight basis 6.0 (23 %), 513 (18 %) and 102 (12 %) pg/g in Veal, Herring and Brown meat, respectively, with total RSDs given in parentheses. The consensus concentrations for PBDE-209 were 12 (n=14, RSD=52 %), 11 (n=21, RSD=111 %) and 14 (n=17, RSD=54 %) pg/g fresh weight in Veal, Herring and Brown meat, respectively. The consensus concentrations calculated for HBCDs are indicative, as only a few laboratories reported results (n=3-6).
9-10 laboratories reported results for the different PFASs in the samples of Perch. The sample was on beforehand known to be heavily contaminated with PFOS (consensus concentration 125393 pg/g fw). 80 % of the participants obtained a Z-score ±2 for PFOS, and 60 % obtained Z-scores ≤1. For PFDA (consensus median 705 pg/g fw), 55 % of the participants obtained z-scores ≤1, and for PFUnDA (consensus concentration was 951 pg/g fw) 56 % obtained z-scores ≤1.
The consensus values for PFHxS, PFOA and PFNA are indicative due to few reported results (many non-detects)