1
Screening of Poly- and Perfluoroalkyl
Substances (PFASs) and Extractable Organic
Fluorine (EOF) in Swedish Blood Samples
Humanexponering av PFAS genom analys av
totalt organiskt fluor (TOF) – en pilotstudie
2
Screening of Poly- and Perfluoroalkyl Substances (PFASs) and Extractable Organic Fluorine (EOF) in Swedish Blood Samples
Humanexponering av PFAS genom analys av totalt organiskt fluor (TOF) – en pilotstudie
Report authors
Rudolf Aro, Örebro University Anna Kärrman, Örebro University Leo Yeung, Örebro University
Responsible publisher Örebro University Postal address
School of Science and Technology Örebro University
S-701 82 Örebro SWEDEN Telephone 019-30 1421 Report title and subtitle
Screening of Poly- and Perfluoroalkyl Substances (PFASs) and Extractable Organic Fluorine (EOF) in Swedish Blood Samples
Purchaser
Swedish Environmental Protection Agency, Environmental Monitoring Unit
SE-106 48 Stockholm, Sweden Keywords for location (specify in Swedish)
Umeå, Stockholm, Uppsala, Örebro, Malmö Keywords for subject (specify in Swedish) blod, PFAS, EOF, novel PFAS, ultrakort PFAS Period in which underlying data were collected 2018
Summary
This report summarises the findings of an investigation into the occurrence of poly- and
perfluoroalkyl substances (PFASs) in the Swedish population. A total of 148 whole blood samples were analyzed in this study from 5 different municipalities in Sweden: Malmö, Stockholm, Umeå, Örebro and Uppsala. Majority of the samples were collected in 2018. Using both liquid and supercritical chromatography coupled with tandem mass spectrometers a total of 63 PFASs were analyzed. These results were then compared with extractable organofluorine (EOF) levels measured with combustion ion chromatography. The data from both target PFAS analysis and EOF was used to perform fluorine mass balance analysis.
In general, the PFAS profile was dominated by long-chain perfluoroalkyl sulfonates (PFSAs with more than 6 fluorinated carbons), on average accounting for 79% of the total PFAS budget. The second most prominent PFAS class were long-chain perfluoroalkyl carboxylates (PFCAs with more than 7 fluorinated carbons), accounting for an additional 19% of the PFAS exposure. The average sum PFAS concentrations from different municipalities in ascending order were 4.6 ng/g – Örebro, 4.7 ng/g – Umeå, 6.0 ng/g – Malmö, 6.3 ng/g – Stockholm and 6.3 ng/g – Uppsala.
On average, the EOF level was 7.8 ng F/g and 63% of the EOF remained unexplained. The average EOF levels from different municipalities in ascending order were 4.95 ng F/g – Uppsala, 5.40 ng F/g – Malmö, 6.67 ng F/g – Örebro, 8.31 ng F/g – Stockholm and 13.4 ng F/g in Umeå.
The most commonly detected novel PFAS was perfluoroethylcyclohexane sulfonic acid (PFECHS), which was found at trace levels in 80% of the samples. Of the ultra-short chain PFASs
perfluoroethane sulfonic acid (PFEtS) was found in 49% of the samples. Trifluoroacetic acid (TFA) was detected in 62% of the samples.
3
Sammanfattning
Denna rapport redovisar resultaten från en studie angående förekomsten av poly- och perfluoroalkylsubstanser (PFAS) i den svenska befolkningen. Totalt har 150 helblodsprover från fem olika kommuner i Sverige (Malmö, Stockholm, Umeå, Uppsala och Örebro) analyserats i studien. Majoriteten av proverna samlades in under 2018. Totalt analyserades 63 PFAS-ämnen genom att använda både vätske- och superkritisk vätskekromatografi kopplad till masspektrometrisk detektion. Dessa resultat jämfördes sedan med de detekterade halterna av extraherbart organisk fluor (EOF) i en massbalans analys av fluor.
PFAS profilen i blodproverna dominerades generellt av långkedjiga perfluoralkyl-sulfonsyror (PFSA med fler än sex fluorerade kol), vilka i genomsnitt stod för 78% av den totala PFAS-halten. Den näst mest framträdande PFAS klassen var långkedjiga perfluoroalkyl-karboxylsyror (PFCA med fler än sju fluorerade kol) som utgjorde ytterligare 20% av PFAS exponeringen. Den genomsnittliga summan av 63 PFAS i från de olika kommunerna i stigande ordning var 4.6 ng/g - Örebro, 4.7 ng/g Umeå, 6.0 ng/g - Malmö, 6.3 ng/g - Stockholm och 6.3 ng/g - Uppsala.
Den genomsnittliga summan av EOF halten var 7.8 ng F/g, vilket betyder att i genomsnitt 63% av EOF var oidentifierade ämnen. De genomsnittliga EOF-nivåerna från de olika kommunerna i stigande ordning var 4.95 ng F/g - Uppsala, 5.40 ng F/g - Malmö, 6.67 ng F/g - Örebro, 8.31 ng F/g - Stockholm och 13.4 ng F/g - Umeå .
Bland de nyare PFAS-ämnena som vanligtvis inte studeras i människor var perfluoroetylcyklohexansulfonsyra (PFECHS) den som hittades i flest prover, i 80% av alla prover. Av de ultrakorta PFAS-ämnena hittades perfluoroetansulfonsyra (PFEtS) i 49% av proverna och trifluorättiksyra (TFA) detekterades i 62% av proverna.
4 Summary
1. Frame of the study... 5
2. Background ... 6
3. Samples for PFASs and EOF screening ... 9
4. Analysis and quantification ...10
4.1. Extraction procedure ... 10
4.2. Quantification of target analytes ... 11
4.2.1. Instrumentation ... 11
4.2.2. Standards and calibration ... 11
4.2.3. Limit of detection and quantification... 12
4.2.4. Recoveries, precision and accuracy ... 12
4.3. Quantification of EOF content ... 14
4.3.1. Instrumentation ... 14
4.3.2. Standards and calibration ... 15
4.3.3. Limit of detection ... 15
4.3.4. Precision and accuracy ... 15
4.4. Data treatment ... 15
5. Results ...16
5.1. Overall levels and distribution ... 16
5.3. Stockholm ... 23 5.4. Umeå ... 26 5.5. Örebro ... 29 5.6. Uppsala ... 32 6. Discussion ...35 6.1. Legacy PFAS ... 35 6.2. Precursor compounds ... 36 6.3. Novel PFAS ... 37
6.4. Fluorine mass balance ... 37
7. Findings ...39
8. Conclusions and Future Work ...39
8. Acknowledgements ...41
9. References ...41
Appendix 1. Full list of target PFASs and their abbreviations ...46
Appendix 2. Instrumental parameters for LC-MS/MS ...48
Appendix 3. LOD range for LC-MS/MS ...50
Appendix 4. Average concentration (ng/g) of different PFASs, percentage of sample above LOQ and between LOD and LOQ ...52
5 1. Frame of the study
The objective of this investigation was to screen for legacy and novel PFASs in human blood samples and perform fluorine mass balance analysis to estimate the levels of unknown organofluorines that people are exposed to. As the samples were taken at different locations in Sweden, it will help elucidate if there are any regional differences in overall human exposure to PFASs. The results from this study can be used to guide further allocation of resources to areas with the greatest need for further investigation.
The target analysis of individual PFASs provided homologue profiles for the different municipalities. Those results, combined with values obtained for environmental samples [1], could be used to track down any possible point sources. The fluorine mass balance analysis could be used as a gauge to estimate conceivable future health risks to the general population – high levels of unidentified organofluorine compounds would warrant further investigations. A total of 63 individual PFASs were monitored in this study and divided into the following groups (Appendix 1):
1. Ultra-short chain PFASs
2. Perfluoroalkyl carboxylic acids and sulfonic acids (PFCAs and PFSAs) 3. Precursor PFASs
4. Perfluoroalkyl phosphonic and phosphinic acids (PFPAs/PFPiAs) 5. Novel PFASs
Many of these groups do not meet the criteria of a persistent organic pollutant set by the Stockholm Convention [2]. Ultra-short chain PFASs (PFASs having between 1 to 3 fluorinated carbons) are not bioaccumulative, but they are persistent and high levels of them have been reported [3]. Some of the novel PFASs have been detected in water samples from Sweden, but there is little information regarding their levels in humans.
6 2. Background
Fluorine is the most common halogen on Earth [4]. However, only a few biologically produced organofluorine compounds have been documented [5], which is in contrast with the prevalence of fluorine in anthropogenic substances. Also, natural compounds only contain a single fluorine atom as opposed to many man-made organofluorines [6]. It has been found that fluorination can improve the effectiveness, absorption and half-life of biologically active compounds [7]– [9]. This has resulted in approximately 50% of pesticides [10] (and around 25% of all agrochemicals [11]) and 25% of pharmaceuticals [12] containing a fluorine atom or a fluoroalkyl group. Examples of these compound classes are given in Figure 2-1. While, these groups of compounds do not belong to the class poly- and perfluoroalkyl substances (PFASs), they have the potential to contribute to the EOFexposure.
Figure 2-1. Examples of some fluorinated compounds that are not PFASs: pyrimisulfan – pesticide; fludrocortisone – pharmaceutical; monofluoroacetate – naturally occuring compound.
Highly fluorinated compounds or PFASs contain one or more perfluoroalkyl moiety -CnF2n+1.
These are all anthropogenic substances and have been produced in quantity since the 1950s [13]. As of 2018, nearly 5000 different PFASs were identified and assigned a CAS number [14], examples of their structure a given in Figure 2-2. Due to their useful properties (e.g. chemical stability, lowering surface tension, water and oil repellency), PFASs have found uses in a wide variety of industries from food packaging [15] to metal plating [16]. However, the same properties have led to their ubiquitous presence in the environment [17]–[19] and living organisms [20]–[22]. Additionally, PFASs have been found in human samples from all over the globe, e.g. Australia [23], United States, [24] China [25].
Figure 2-2. Structures of the most commonly detected PFASs; A: perfluorooctane sulfonate (PFOS), B) perfluorooctanoate (PFOA).
The first evidence of organofluorine compounds in human serum were reported at the end of the 1960s and the mid 1970s [26], [27]. The identification and quantification of the individual PFASs became feasible a few decades later, when both sample extraction and instrumental methods caught up with the analytical needs [28], [29]. Coupled with the development of more isotopically labelled standards, it has facilitated more research in this field [30], [31].
7
As more research has been performed, evidence of the negative health effects of PFASs has been mounting. These compounds have been linked to adverse effects on the immune system [32], development [33] and suspected to have detrimental effects on human fertility [34], [35]. These and environmental concerns have led to regulation of some PFASs – perfluorooctane sulfonate (PFOS) and perfluorooctanoate (PFOA) are included in the Stockholm Convention’s list of persistent organic pollutants (POPs) with perfluorohexane sulfonate (PFHxS) under review for its inclusion [36], [37].
While such efforts are a step towards a better understanding of PFAS exposure, it is still far from a complete picture. The number of PFASs being monitored even in broader frameworks, such as the 11 PFASs in Swedish drinking water [38]. More strict regulation has been met with diversification on the side of producers, shifting to polyfluorinated phosphate esters (PAPs), shorter-chain PFASs (compounds with less than 6 fluorinated carbon for sulfonates and less than 7 fluorinated carbons for carboxlyates) and polyfluorinated ether acids. Examples of “novel” PFASs are given in Table 2-1. In addition to the thousands of PFASs being produced, there are precursor [39] and intermediate [40] compounds. It is necessary to estimate the combined exposure to as many different organofluorine compounds as possible.
Table 2-1. Novel PFAS included in this study.
Name Abbreviation CAS nr.
Replace-ment for Structure 6:2 chlorinated polyfluorinated ether sulfonate 6:2 Cl-PFESA F-53B (major) 73606-19-6 PFOS F F F F F F F F F F F F O F F Cl F F S O O -O K+ 8:2 chlorinated polyfluorinated ether sulfonate 8:2 Cl-PFESA F-53B (minor) PFOS F F F F F F F F F F F F O F F F F S O O -O F F Cl F F K+ Perfluoro-4- ethylcyclohexane-sulfonate PFECHS 335-24-0 S O O -O F F F F F F F F F F F F F F F K+ 3H-perfluoro-3-[(3-methoxy-propoxy) propanoic acid] ADONA 958445-44-8 PFOA O O O O -F F F F F F F F F F F F H NH4 + Hexafluoropropylene oxide dimer acid
HFPO-DA (also known as GenX) 62037-80-3 PFOA O O -F F F F O F F F F F F F NH4+
There are several approaches to achieve this goal. One of them is the use of total oxidizable precursor assay (TOPA), a method based on oxidation of precursor compounds into stable and measurable perfluoroalkyl acids (PFAAs) [41]. These PFAAs are readily analysed using liquid chromatography coupled with tandem mass spectrometers (LC-MS/MS) with suitable mass labelled standards. The concentration of PFAAs is measured before and after oxidation, with the change attributed to precursor compounds. This approach has been used in several studies [42], [43], but the method development can prove time-consuming and is associated with some problems [44].
8
Another method is to use high-resolution mass spectrometry (HR-MS) coupled with a suitable separation method (e.g. liquid chromatography) to identify novel PFASs. This has been applied to environmental samples using liquid chromatography coupled with quadrupole time of flight instrumentation (LC-qTOF) [45]. However, confirmation of an identified structure requires a reference standard or at least a library match [46], making data analysis time-consuming.
Figure 2-3. Scheme showing the different types of fluorine present in a samples.
The fluorine mass balance approach can be an option to estimate the amount of unidentified organofluorine in a sample. The total fluorine (TF) content of a sample is made up of both inorganic fluorine (IF) and organic fluorine (OF, dark blue in Figure 2-3). As the IF levels can be an order of magnitude higher than that of OF [47] and the combustion ion chromatography (CIC) method does not distinguish between IF and OF, it is important to separate these two types of fluorine. This is usually achieved with an appropriate sample extraction, which can also help to reduce interferences and improve detection limits. However, the OF present in a sample is further divided into non-extractable organic fluorine (NEOF) and extractable organic fluorine (EOF). Depending on the chosen extraction method, some organofluorine compounds might not be extracted from the sample – forming the NEOF fraction. The organofluorine compounds that are extracted constitute the EOF (light blue in Figure 2-3) – which is analysed for both target PFAS and fluorine content for fluorine mass balance analysis. The amount of identified organic fluorine (blue pattern in Figure 2-3) is calculated from the levels of target PFASs, using the formula presented by Figure 2-4. This conversion has to be done for each target compound separately as the degrees of fluorination and molecular weights are different. This value is substracted from the measured EOF content to find the fraction that remains unidentified – the unidentified organic fluorine (UOF). The fluorine mass balance approach has been applied to various matrices: blood [47]–[49], water [1], [50] and various biota samples [1].
Figure 2-4. Formula for converting from PFASs to fluorine. CF: the corresponding fluoride concentration
(ng×F×g-1); nF: the number of fluorine atoms in the PFAS molecule; MWF: the molecular weight of fluorine;
MWPFAS: the molecular weight of the individual target PFAS; CPFAS: the concentration of the target PFAS from
9 3. Samples for PFASs and EOF screening
The partitioning of many PFASs between whole blood and plasma/serum is not known. Additionally, a recent study showed that some compounds are preferentially found in only one sample matrix (e.g. PFHxA in whole blood) [51], probably due to preferential binding of some PFASs to blood proteins, as the cellular materials are removed in plasma/serum. The partitioning of EOF between between whole blood, serum and plasma has not been studied in detail. There is a risk that some organofluorine compounds may be left behind when separating serum or plasma from whole blood. Thus whole blood was the matrix of choice for this investigation to represent human exposure to PFASs and EOF.
The whole blood samples were collected when donors went to donate blood at different blood centres. The study was approved by the Ethics Committee in Uppsala and the participants gave a written informed consent. Between 4 and 9 mL of whole blood was collected from each individual into one or several vacutainer tubes. These containers were in turn stored at +4 °C, shipping was done in thermally isolated boxes to avoid excessive heating. Once in at MTM the samples were stored at +4 °C again. Each sample was coded and only the gender, age and sampling location data were retained. When the samples were received by the laboratory, they were assigned a local code for internal use. Of the 150 volunteers, 51 % were female and 49 % were male; the age ranged from 11 to 97 years old with a median age of 52 years. Initially, samples were acquired from 6 municipalities; however, blood samples received from Linköping were plasma. The plasma samples were excluded from this investigation as they are not directly comparable to whole blood samples, locations and the numbers of samples for the remaining municipalities are presented by Figure 3-1 and Table 3-1.
Figure 3-1. Sampling locations for the whole blood.
Table 3-1. Number of samples from each location.
Location Number of samples
Malmö 30
Stockholm 30
Umeå 30
Örebro 30
10 4. Analysis and quantification
All samples were extracted in duplicate (see Figure 4-1), the first one (Replicate 1) was spiked with internal standards (IS) before the extraction and used for target analysis (more details in Section 4.1 and Figure 4-2). The second replicate (Replicate 2) was extracted without spiking any IS and analyzed for EOF content (more details in Section 4.1 and Figure 4-2).
Figure 4-1. Overview of the sample analysis scheme.
4.1. Extraction procedure
Prior to sample extraction, individual whole blood samples were vigorously shaken and/or vortexed to mix the contents of each vacutainer. Two aliquots of the whole blood were taken into pre-cleaned 15 mL polypropylene (PP) tubes (rinsed three times with methanol (MeOH) and allowed to dry), the mass of each sub-sample was recorded. The subsample for target analysis (Replicate 1) was spiked with an IS mixture; the second subsample, for EOF analysis (Replicate 2), was extracted without any IS. The omission of IS for Replicate 2 was necessary as this would interfere with the EOF analysis, because the CIC system cannot differentiate between different sources of fluorine. These duplicate samples were extracted in the same batch to minimize the variability between them.
Samples were extracted in duplicates using the ion pair method [52]. In brief, 2 mL of 0.5 M tetrabutyl-ammonium (TBA) solution in water was added to the extract. Then, 5 mL of methyl tert-butyl ether (MTBE) was added to the tube. The mixture was shaken horizontally for 15 minutes at 250 rpm and centrifuged for 10 minutes at 8000 g to separate the organic and aqueous phases. The top layer (MTBE) was transferred to a new pre-cleaned PP tube and the extraction was repeated twice with 3 mL of MTBE. The extracts were combined and evaporated to 200 μL using an evaporation system. The combined extracts were reconstituted to 1.0 mL with MeOH and evaporated 0.5 mL with the evaporation system and the supernatants were transferred to LC vials.
The sample extracts were then split for different instrumental analyses as shown in Figures 4-1 and 4-2-1. Most of the analytes were quantified in the sample with 40% organic solvent content. The sample with 80% organic solvent content was used for PAPs and ultra-short chain PFAS analyses.
11
Figure 4-2-1. Detailed scheme of how the samples were divided for different instrumental analysis. Replicate 1 was analyzed with two different methanol compositions, 40% and 80% to improve chromatography. Replicate 2 was analyzed for the EOF content with CIC. RS – mass labelled recovery standard; aqueous phase – 2 mmol/L ammonia acetate in MilliQ water; all extracts were in methanol (MeOH).
4.2. Quantification of target analytes 4.2.1. Instrumentation
Analytes with four or more fluorinated carbons were quantified by ultra performance liquid chromatography electrospray ionization tandem mass spectrometry (UPLC-ESI-MS/MS) in negative mode. The analytes were separated on a Waters Acquity UPLC with a BEH column (2.1 × 100 mm, 1.7 μm) coupled to a Waters XEVO TQ-S MS/MS. The mobile phases were methanol (MeOH) and 30:70 MeOH:MilliQ water mixture, both with 2 mmol/L ammonium acetate and 5 mmol/L 1-methylpiperidine as additives [53]. Ultra-short chain compounds (C2-C3) were separated by a supercritical fluid chromatographic system (Waters Ultra Performance Convergence Chromatograph, UPCC), using CO₂ and MeOH with 0.1% ammonia as mobile phases with a Torus DIOL analytical column (3.0 × 100 mm, 1.7 μm). The UPCC was coupled to the Waters XEVO TQ-S detector [3]. Levels of two novel compounds (HFPO-DA and ADONA) were monitored using a Waters Acquity UPLC system coupled with a XEVO TQ-S micro MS/MS, the mobile phases and column were as described above. The transitions monitored by mass spectrometers are given in Appendix 2.
4.2.2. Standards and calibration
Quantification of the analytes was done using native and isotope labelled internal standards purchased from Wellington Laboratories (Guelph, Canada), except for 10:2 monoPAP and 10:2 diPAP, which were purchased from Chiron (Trondheim, Norway). The PFOS isomers were reported as a sum of individual isomers, 1-m-PFOS, 6/2-m-PFOS, 3/4/5-m-PFOS, 4.4/4.5/5.5-m2-PFOS. Concentrations of all analytes were recovery-corrected using labelled
internal standards. For those homologues of PFCAs, PFSAs, PAPs, fluorotelomer sulfonates (FTSAs), fluorotelomer carboxylates/fluorotelomer unsaturated carboxylates
12
(FTCA/FTUCAs), and perfluoroalkyl sulfonamidoacetates (FOSAAs) where no isotope labelled standard were available, the internal standard closest in retention time within the same compound class was used for quantification. For Cl-PFESAs, PFECHS, polyfluorinated ether carboxylates (PFECAs), perfluoroalkyl phosphonates/ perfluoroalkyl phosphinates (PFPA/PFPiAs), and ADONA, the IS closest in retention time of the compound classes PFCAs and PFSAs was used for quantification. Multiple reaction monitoring (MRM) was used and at least two transitions were monitored for all analytes, except for TFA, perfluoropropanoate (PFPrA), perfluorobutanoate (PFBA) and perfluoropentanoate (PFPeA) where one transition was monitored. Due to poor recoveries of PFPrA and TFA in the blood samples, their concentrations were not reported but their detection are indicated.
In total, the levels of 63 PFASs were monitored in this study. The choice of analytes was based on previous studies and aimed to cover the most commonly found PFASs – PFAAs, to which people have had historical exposure and which are the stable degradation products of different precursors compounds. While PFAAs have accounted for most of the known PFASs exposure, several classes of PFCA and PFSA precursors were included in an attempt to elucidate possible exposure pathways to PFASs. Several intermediates (e.g., FTCAs and FTUCAs) were also monitored to assess human exposure to precursors. Besides, some ultrashort and novel PFASs (e.g., ADONA, GenX,) were also included to assess human exposure, as their information is limited.
The concentrations of each analyte were calculated using relative response factors (RRF, see Figure 4-2-2-1). The RRF was determined by analyzing calibration samples containing both the native (¹²C) and isotope labelled (¹³C) compounds. The calibration range was from approximately 0.005 to 30 ng/mL, the limit of detection (LOD) of each analyte is given in Appendix 3.
Figure 4-2-2-1. Calculation of analyte concentration in a sample; Cx - analyte concentration, CIS - internal
standard concentration, Ax - peak area of analyte, AIS - peak area of internal standard, RRF - relative response
factor determined separately.
4.2.3. Limit of detection and quantification
The limit of detection (LOD) for target analytes was determined separately for each sample preparation batch, it was calculated as the sum of the procedural blank and three times the pooled standard deviation of the analyte. The limit of quantification (LOQ) was determined as the procedural blank plus 10 times the pooled standard deviation. If a compound was not found in any of the procedural blanks, the lowest point of the calibration curve was used as the LOQ instead.
4.2.4. Recoveries, precision and accuracy
Samples with recoveries between 20 and 150 % were considered acceptable and the analyte concentrations were calculated using internal standards. The recoveries for different internal standards are given in Table 4-2-4-1. Samples with IS recoveries below 20% or great than 150% were not reported and were denoted as not quantified (n.q.) in the results.
Each extraction batch included a quality control (QC) samples to monitor both accuracy and reproducibility. The QC sample was the Standard Reference Material (SRM) 1957, organic contaminants in non-fortified human serum (National Institute of Standards and Technology
13
(NIST); Maryland, United States). The observed relative standard deviations (RSD) of L-PFOS (linear PFOS) and L-PFOA (linear PFOA) concentrations in QC samples were below 20%. Additional spike-recovery experiments were done with pooled whole blood, a selection of the results (for novel PFASs) is presented in Table 4-2-4-2.
14
Table 4-2-4-1. Results of internal standard recovery with relative standard deviation (RSD) and the number of samples where the recovery was within the acceptable range (20-150 %).
Analyte Average recovery RSD n 13C-PFBA 42% 34% 149 13C-PFPeA 47% 33% 148 13C-PFHxA 47% 30% 149 13C-PFHpA 61% 35% 149 13C-PFOA 55% 27% 149 13C-PFNA 55% 27% 148 13C-PFDA 54% 25% 146 13C-PFUnDA 57% 33% 141 13C-PFBS 91% 29% 139 18O-PFHxS 85% 17% 150 13C-PFOS 78% 21% 149 13C-6:2 monoPAP 62% 30% 115 13C-8:2 monoPAP 70% 32% 116 13C-6:2 diPAP 35% 33% 118 13C-8:2 diPAP 48% 45% 94 2H -EtFOSAA 49% 32% 136 13C-HFPO-DA 52% 36% 150
Table 4-2-4-2. Results of spike-recovery experiments (%) for novel PFASs (4 ng) in blood.
Analyte Average recovery n
6:2 Cl-PFESA 86%. 2
8:2 Cl-PFESA 103% 2
PFECHS 87% 2
ADONA 61% 2
HFPO-DA 58% 2
4.3. Quantification of EOF content 4.3.1. Instrumentation
Extractable organofluorine (EOF) content was measured using a combustion ion chromatography (CIC) system. The CIC consists of a combustion module (Analytik Jena, Germany), a 920 Absorber Module and a 930 Compact IC Flex ion chromatograph (both from Metrohm, Switzerland). Separation of anions was performed on an ion exchange column (Metrosep A Supp 5 – 150/4.0) using carbonate buffer (64 mmol/L sodium carbonate and 20 mmol/L sodium bicarbonate) as eluent for isocratic elution. In brief, the sample extract (0.1 mL) was injected on to a quartz boat, which was pushed into the furnace by the
15
autosampler. The furnace was kept at 1000-1050 °C for combustion, during which, all organofluorine compounds were converted into hydrogen fluoride (HF). A carrier gas (argon) was constantly pumped through the combustion tube, the gas carries all formed HF into the absorber module where MilliQ water is used to capture the HF. A 2 mL aliquot of the absorber solution is then injected on a pre-concentration column and then injected on the ion chromatograph. The concentration of F¯ ions in the solution was measured using ion chromatography.
4.3.2. Standards and calibration
Standard solutions from a solid PFOS potassium salt (Fluka, part of Fisher Scientific, Hampton, United States) were prepared in methanol. These solutions were used to created injection standards to monitor the performance of the CIC system. Quantification of samples was based on an external calibration curve. For both calibration and project samples the peak area of the preceding combustion blank was subtracted from peak area of the sample to correct for the background contamination. A five point calibration curve (50, 100, 200, 500 and 1000 ng F/mL) was constructed, with each level analyzed in triplicates.
Fluoride signal was observed in combustion blank even when no sample was analyzed. Prior to sample analysis, multiple combustion blanks were performed until stable fluoride signals were reached; the RSD of the three most recent combustion blanks lower than 5 %.
4.3.3. Limit of detection
The limit of detection (LOD) was determined separately for each sample preparation batch, the procedural blank of the batch plus three times the pooled standard deviation of the procedural blanks. The reported values were not corrected for extraction blanks.
4.3.4. Precision and accuracy
Combustion blanks (CIC analysis cycle without a sample) were made between sample injections to evaluate the presence of carryover between samples and to obtain a reliable estimate of the background fluorine levels. The repeatability of the instrument was tested by triplicate analysis of dilutions made from an anion SRM solution (product code 89886, Sigma-Aldrich). The five dilutions were in the range of 60 ng F/g to 1200 ng F/g and the relative standard deviation at all five dilution levels was below 25%. The calibration curve, which was made from a PFOS salt by a series of dilutions by weight, was compared to an older calibration curve. The difference of the slope of these two separate calibrations was below 10%. This was in the same range as the relative standard deviation of the calibration point replicates.
4.4. Data treatment
When concentrations of analytes were below LOQ, zero was assigned for them for any further data treatment; this approach was used for both target PFAS and EOF data. Thus, when calculating the sum concentration of the 63 PFASs (∑₆₃PFAS) in a sample, the concentrations of individual analytes were added up with those below LOQ were kept as zero. Similarly, for the calculation of PFAS profiles for municipalities, the average concentration of each analyte was calculated first and when the compound was below LOQ, zero was assigned for the calculation. When calculating detection frequencies for analytes, samples with levels above LOD were used.
16 5. Results
5.1. Overall levels and distribution
A total of 150 whole blood samples from Malmö, Stockholm, Umeå, Örebro and Uppsala were analysed for their PFAS content. Two samples were excluded from further data analysis as PFOS (one of the major compounds) could not be reliably quantified. The average sum concentration of the 63 PFASs monitored in this study (∑63PFAS) for the 148 samples was
5.6 ng/g. Detailed average sum PFAS concentrations of different classes and their respective detection frequencies in different municipalities are provided in Table 5-1. Details of average concentrations of different individual PFAS and the percentages of samples above respective LOD and LOQ are provided in Appendix 4.
Of the 63 analytes monitored in this study, 40 were detected at least once. PFOA, PFNA, PFHxS, perfluoroheptane sulfonate (PFHpS), and PFOS had a detection frequency above 99 %; with 71% of detection of PFDA and 49% for PFUnDA. The average PFAS concentrations in descending order are given as follows: PFOS (L-PFOS: 1.59 ng/g, 3/4/5-m-PFOS: 1.36 ng/g, 6/2-m-PFOS: 0.41 ng/g, dimethyl-PFOS: 0.11, and 1-m-PFOS: 0.09 ng/g), PFHxS (0.59 ng/g), PFOA (0.56 ng/g), PFNA (0.27 ng/g), PFDA (0.13 ng/g), and PFUnDA (0.08 ng/g). The average ∑63PFAS levels from the five municipalities were as follows: 6.3 ng/g (0.9 – 22.0 ng/g)
in Uppsala, 6.3 ng/g (1.9 – 20.0 ng/g) in Stockholm, 6.0 ng/g (2.0 – 20.0 ng/g) in Malmö, 4.7 ng/g (1.2 – 13.0 ng/g) in Umeå and 4.6 ng/g (0.4 – 9.9 ng/g) in Örebro (Figure 5-1-1).
The PFAS homologue profiles were similar among the five municipalities; long-chain PFSAs and long-chain PFCAs were dominating the total PFAS exposure. The former accounted for between 77 and 81 % of the ∑63PFAS while the long-chain PFCAs made up an additional 16
to 22 % of ∑63PFAS. Within long-chain PFSAs, PFOS was dominant (L-PFOS - 28 % of the
∑63PFAS on average, 3/4/5-m-PFOS - 25 %), followed by PFHxS (10 %). The most abundant
long-chain PFCAs were PFOA (10 % of the ∑63PFAS), perfluorononanoate (PFNA, 5 %) and
perfluorodecanoate (PFDA, 2 %). The remaining PFAS classes (short-chain PFCAs and PFSAs, ultra-short chain PFSAs, PFCA and PFSA precursors and novel PFASs) each contributed up to 0.8 % to the ∑63PFAS. A more detailed overview of the homologue profiles
is presented by Figure 5-1-2.
Six PFASs (PFOA, PFNA, PFDA, PFHxS, PFHpS and PFOS) had detection frequencies above 90 %. A further three compounds (TFA, PFECHS and perfluoropentane sulfonate (PFPeS)) were detected in more than half of the samples. PFCA precursors detected were 6:2 FTSA and 8:2 FTSA; they were detected in 16 % and 24 % of the samples respectively. Of the PFSA precursors, N-methyl perfluorooctane sulfonamidoacetate (MeFOSAA) and perfluorooctane sulfonamide (FOSA) were found in 16 % and 15 % of the samples respectively.
17
Table 5.1 a) Average sum PFAS concentrations (ng/g whole blood) of different classes and their respective b) contribution (%) to the average sum PFAS in different municipalities
a) Malmö Stockholm Umeå Örebro Uppsala Overall
n=29 n=29 n=30 n=30 n=30 n=148 Concentration (ng/g) ∑ultrashort 0.02 0.01 0.00 0.01 0.01 0.01 ∑PFCA 1.03 0.96 0.97 1.37 1.13 1.09 ∑PFSA 5.15 3.64 3.71 4.86 4.86 4.45 ∑FTSA 0.01 0.01 0.01 0.00 0.00 0.00 ∑FTCA 0.00 0.00 0.00 0.00 0.00 0.00 ∑FTUCA 0.01 0.00 0.00 0.00 0.00 0.00 ∑FASA/FASE 0.01 0.01 0.00 0.00 0.00 0.00 ∑FOSAA 0.02 0.05 0.01 0.00 0.02 0.02 ∑PAP 0.00 0.00 0.00 0.00 0.00 0.00 ∑SamPAP 0.00 0.00 0.00 0.00 0.00 0.00 ∑PFPA 0.00 0.00 0.00 0.00 0.00 0.00 ∑PFPiA 0.00 0.00 0.00 0.00 0.00 0.00 ∑Novel 0.03 0.02 0.04 0.03 0.02 0.03 Total 6.28 4.70 4.74 6.28 6.04 5.61
b) Malmö Stockholm Umeå Örebro Uppsala Overall
Composition (%) ∑ultrashort 0.3% 0.2% 0.1% 0.2% 0.2% 0.2% ∑PFCA 16.5% 20.4% 20.5% 21.8% 18.7% 19.5% ∑PFSA 82.1% 77.6% 78.2% 77.4% 80.4% 79.3% ∑FTSA 0.1% 0.1% 0.2% 0.0% 0.0% 0.1% ∑FTCA 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% ∑FTUCA 0.1% 0.0% 0.0% 0.0% 0.0% 0.0% ∑FASA 0.1% 0.2% 0.0% 0.1% 0.0% 0.1% ∑FOSAA 0.4% 1.0% 0.2% 0.0% 0.3% 0.4% ∑PAP 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% ∑SamPAP 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% ∑PFPA 0.1% 0.0% 0.0% 0.0% 0.0% 0.0% ∑PFPiA 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% ∑Novel 0.4% 0.5% 0.8% 0.5% 0.3% 0.5%
∑ultrashort - PFEtS, PFPrS; TFA and PFPrA were not quantified
∑PFCA - PFBA, PFPeA, PFHxA, PFHpA, PFOA, PFNA, PFDA, PFUnDA, PFDoDA, PFTrDA, PFTDA, PFHxDA, PFOcDA
∑PFSA - PFBS, PFPeS, PFHxS, PFHpS, PFOS, PFNS, PFDS, PFDoDS ∑FTSA - 4:2 FTSA, 6:2 FTSA, 8:2 FTSA, 10:2 FTSA
∑FTCA - 3:3 FTCA, 5:3 FTCA, 7:3 FTCA
∑FTUCA - 6:2 FTUCA, 8:2 FTUCA, 10:2 FTUCA ∑FASA - FBSA, MeFBSA, FHxSA, MeFHxSA, FOSA ∑FOSAA - FOSAA, MeFOSAA, EtFOSAA
∑PAP - 6:2 mPAP, 8:2 mPAP, 10:2 mPAP, 6:2 diPAP, 6:2/8:2 diPAP, 8:2 diPAP, 10:2 diPAP ∑SamPAP - SAmPAP, diSAmPAP
∑PFPA - PFHxPA, PFOPA, PFDPA ∑PFPiA- 6:6 PFPiA, 6:8 PFPiA, 8:8 PFPiA
18
Figure 5-1-1. Average concentration of different PFAS classes in the whole blood from different Swedish municipalities.
19
Fluorine mass balance analysis was done for 147 samples because 1 sample there was not enough blood for duplicate extraction. Of those 147 samples, 88 % had EOF levels above the limit of detection, ranging from 0.51 to 48.7 ng F/g with an average of 7.8 ng F/g (Figure 5-1-3). The average EOF levels for different municipalities were 13.4 ng F/g in Umeå, 8.31 ng F/g in Stockholm, 6.67 ng F/g in Örebro, 5.40 ng F/g in Malmö and 4.95 ng F/g in Uppsala. The largest fraction EOF was explained (indentified PFAS, iPFAS) in the samples from Uppsala – 57 %, followed by Malmö (52%) and Stockholm (36%). The fraction of iPFAS was lower in Örebro (25 %). The smallest iPFAS fraction was in the samples from Umeå (15 %).
Figure 5-1-3. Fluorine mass balance graphs for the five municipalities on the left, where the white bars (iPFAS is the sum of targeted PFASs and the black bars are unidentified organofluorines (UOF). To the right are the experimental results with the average EOF concentrations per town, the range (min and max values) and the number of samples with EOF levels above LOD.
20 5.2. Malmö
The PFAS homologue profiles of the samples from Malmö were dominated by long-chain PFSAs and PFCAs, on average accounting for 80% and 18% of ∑63PFAS (Figure 5-2-1). The
most abundant long-chain PFSAs were PFOS (L-PFOS - 30% of ∑63PFAS, 3/4/5-m-PFOS -
26%, 6/2-m-PFOS - 8.7%), followed by PFHxS (8.5%) and PFHpS (5.0%). These PFSAs were detected in all samples from Malmö. Short-chain PFSAs (PFBS and PFPeS) made up only 0.1% of the ∑63PFAS. Of the short-chain PFSAs, perfluorobutane sulfonate (PFBS) was present in
33% of samples and PFPeS only in 7% of the samples. Perfluoroethane sulfonate (PFEtS) was the one detected of the ultra-short chain PFSAs, with a maximum concentration of 0.04 ng/g, which was found in 60% of the samples and on average in made up 0.2% of the ∑₆₃PFAS. The most abundant long-chain PFCAs were PFOA and PFNA, which contributed 11% and 5.4% to the ∑63PFAS respectively. Smaller contributions also came from PFDA (1.7% of
∑63PFAS) and perfluoroundecanoate (PFUnDA,0.4%). PFOA and PFNA were found in all
samples from Malmö. The PFCAs with longer perfluorinated carbon backbones showed lower detection frequencies (e.g. PFDA in 83% and PFUnDA in 17% of the samples). The short-chain PFCAs (PFBA and PFPeA) together on average made up 0.2% of ∑63PFAS. PFPeA was found
in 33% of the samples and PFBA in 7% while PFHxA was not detected in any samples from Malmö. For ultra-short chain PFCAs, TFA and PFPrA, were detected in 20% and 17% of the samples.
The PFCA and PFSA precursors made up 0.01% and 0.3% of ∑63PFAS respectively. The most
commonly found PFCA precursor were 8:2 FTSA and 6:2 mPAP, both found in 6.7% of the samples. The most commonly detected PFSA precursors were MeFOSAA (found in 20.0% of the samples) and FOSA (in 10.0% of samples).
The most commonly detected novel PFAS was PFECHS, which was detected in all samples from Malmö. The average PFECHS concentration was 0.017 ng/g, ranging from 0.004 ng/g to 0.044 ng/g. Of the two PFESAs, only the 6:2 Cl-PFESA was found, that in two samples – nr 5 (0.004 ng/g) and 18 (0.010 ng/g). Trace levels of ADONA were detected in samples nr. 16 and 22.
For the samples from Malmö, 29 out of 30 had EOF levels above the LOD (Figure 5-2-2). Their average EOF concentration was 5.40 ng F/g and on average 52% of it was accounted for by the 63 PFASs monitored in this study (identified PFAS, iPFAS), more details in Figure 5-2-2. Samples nr. 2, 4, 10, 12, 17, 21, 25 and 28 had all of their EOF explained by the iPFAS. The three samples with the highest EOF (nr. 22, 6 and 9) had a higher than average proportion of unidentified organofluorine compounds (UOF), 75%.
21
22
23
5.3. Stockholm
The PFAS homologue profiles of the samples from Stockholm were dominated by long-chain PFSAs and PFCAs, on average making up 77% and 22% of ∑63PFAS (Figure 5-3-1). The
long-chain PFSAs were dominated by PFOS (L-PFOS - 31% of ∑63PFAS, 3/4/5-m-PFOS - 18%,
dimethyl-PFOS - 6.6% and 6/2-m-PFOS - 5.9%), followed by PFHxS (9.1%). Of the long-chain PFSAs, L-PFOS, 6/2-m-PFOS, 3/4/5-m-PFOS, PFHpS and PFHxS were found in all samples from Stockholm. Both short-chain PFSAs, PFBS and PFPeS, made up of 0.1% of ∑63PFAS on
average. PFPeS was detected more frequently (60%) than PFBS (47%). Only PFEtS of the ultra-short chain PFSAs was found in samples from Stockholm, on average making up 0.2% of ∑63PFAS; this compound was detected in 50% of the samples with a maximum concentration
of 0.03 ng/g.
The long-chain PFCAs that made the highest contribution to the ∑63PFAS were PFOA (11% of
∑63PFAS), PFNA (5.8%), PFDA (2.7%) and PFUnDA (2.1%). The detection frequencies
decreased with increasing perfluorinated carbon chain length. PFOA and PFNA were found in all samples from Stockholm, while PFDA and PFUnDA were found in 93% and 73% of samples respectively. Of the short-chain PFCAs, PFBA contributed 0.2% to the ∑63PFAS; it was found
in 7% of the samples from Stockholm. The ultra-short chain PFCAs, TFA and PFPrA, were detected in 67% and 3% of the samples respectively.
The PFCA and PFSA precursors combined only made up 0.1% of ∑63PFAS, most of which
came from PFSA precursors FOSA, MeFOSAA and perfluorobutane sulfonamide (FBSA). These compounds had the highest detection frequencies as well 13%, 6.7% and 20% respectively. From the 23 PFCA precursors only 6:2 FTSA and 8:2 FTSA were detected in more than 5% of the samples – trace levels were found in 6.7% and 13% of samples respectively.
The most commonly detected novel PFAS was PFECHS, which was detected in 90% of the samples from Stockholm. The average PFECHS concentration was 0.028 ng/g, with a maximum of 0.10 ng/g. The second most common novel PFAS was ADONA, which was detected in 13 samples, with an average concentration of 0.01 ng/g. The third novel compound that was found in samples from Stockholm was 6:2 Cl-PFESA. It was found in 9 samples with an average concentration of 0.002 ng/g.
Of the samples collected from Stockholm, 90% had EOF levels above the LOD (Figure 5-3-2). Their average EOF concentration was 8.31 ng F/g and the known PFAS (iPFAS) accounted for 36% indicating the remaining 64% being unidentified. Samples nr. 6, 21, 22 and 28 had all of their EOF explained by the PFASs monitored in this study. The samples with the highest EOF concentration, nr. 10, 30 and 24, had a higher than average fraction of UOF – 91%.
24
25
26
5.4. Umeå
The PFAS homologue profiles of the samples from Umeå were dominated by long-chain PFSAs and PFCAs, on average making up 78% and 20% of ∑63PFAS (see Figure 5-4-1). The
long-chain PFSAs were dominated by PFOS (L-PFOS - 25% of ∑₆₃PFAS, 3/4/5-m-PFOS - 32%, 6/2-m-PFOS - 7.9%), followed by PFHpS (6.9%) and PFHxS (4.4%); all of these compounds were found in all samples from Umeå with the exception of PFHpS, which was detected in 97% of the samples. Both short-chain PFSAs, PFBS and PFPeS made comparitevly low contributions towards the overall PFAS budget, on average accounting for 0.05% and 0.08% of the ∑63PFAS. PFBS was detected in 23% of the samples and PFPeS in 50%. Only PFEtS of the
ultra-short chain PFSAs was found in samples from Umeå, on average making up 0.1% of ∑63PFAS; this compound was detected in 27% of the samples with a maximum concentration
of 0.02 ng/g.
The long-chain PFCAs that made the highest contribution towards the overall PFAS budget were PFOA (9.3% of ∑₆₃PFAS), PFNA (4.5%), PFDA (2.9%), PFUnDA (1.8%) and perfluorododecanoate (PFDoDA, 1.1%). The detection frequencies decreased with increasing perfluorinated carbon chain length – PFOA and PFNA were found in all samples from Umeå, while PFDA, PFUnDA and PFDoDA were found in 97%, 50% and 27% of samples, respectively. The short-chain PFCAs together made up 0.002% of ∑63PFAS; PFPeA and
perfluorohexanoate (PFHxA) were found at trace levels in 3% of the samples. The ultra-short chain PFCAs, TFA and PFPrA, were detected in 73% and 57% of the samples respectively. The PFCA and PFSA precursors made up 0.4% of ∑63PFAS. The most abundant PFCA
precursors was 8:2 FTSA, which made up 0.1% of ∑63PFAS. The most commonly detected
PFCA precursors were 8:2 FTSA, 6:2 FTSA and 10:2 FTUCA, which were found in 43%, 23% and 6.7% of the samples from Umeå. Of the PFSA precursors, FOSAA and MeFOSAA both contributed 0.1% to the ∑₆₃PFAS; they were detected in 3.3% and 13% of the samples. Other detectable PFSA precursors were FBSA and FOSA, found in 17% and 10% of the samples respectively.
The most commonly detected novel PFAS was PFECHS, which was detected in 15 samples from Umeå. The average PFECHS concentration was 0.01 ng/g, with a maximum of 0.05 ng/g. The second most common novel PFAS was ADONA, which was detected in 4 samples, with a maximum concentration of 0.03 ng/g.
From the 30 samples collected in Umeå, 27 had EOF levels above the LOD (Figure 5-4-2). The average EOF concentration was 13.4 ng F/g and the identified PFAS (iPFAS) accounted for 15% of it and the remaining 85% was unidentified. The samples nr. 12, 17 and 23 had all of thier EOF explained by the target PFAS. Sample nr. 26 had the highest fraction of unidentified organofluorines – 97%, followed by samples nr. 25 and 29 which had 93% of their EOF unexplained.
27
28
29
5.5. Örebro
The PFAS homologue profiles of the samples from Örebro were dominated by long-chain PFSAs and PFCAs, on average making up 77% and 20% of ∑63PFAS (Figure 5-5-1). The
long-chain PFSAs were dominanted by PFOS (L-PFOS - 32% of ∑63PFAS, 3/4/5-m-PFOS - 25%,
6/2-m-PFOS - 7.1%), followed by PFHxS (6.8%) and PFHpS (4.4%); these compounds were found in all samples from Örebro. Both short-chain PFSAs, PFBS and PFPeS made comparatively low contributions towards the overall PFAS budget, on average both accounting for 0.1% of the ∑₆₃PFAS. PFBS was detected in 17% of the samples and PFPeS was detected in 59%. Of the ultra-short chain PFSAs, PFEtS was found in 31% of the samples (maximum concentration of 0.02 ng/g), on average making up 0.1% of ∑63PFAS; PFPrS was found in only
one sample (0.09 ng/g).
The long-chain PFCAs that made the highest contribution to the ∑63PFAS were PFOA (10% of
∑63PFAS), PFNA (4.7%), PFDA (2.9%) and PFUnDA (1.3%). The detection frequencies
decreased with increasing perfluorinated carbon chain length. PFOA and PFNA were found in 97% of samples from Örebro, while PFDA and PFUnDA were found in 93% and 38% of the samples, respectively. The short-chain PFCAs were found only in one sample and together made up 0.2% of ∑63PFAS. The ultra-short chain PFCAs, TFA and PFPrA, were detected in
69% and 28% of the samples respectively.
The PFCA and PFSA precursors combined made up 1.3% of ∑63PFAS. The most abundant
PFCA precursors were 6:2 FTSA and 8:2 FTSA, both accounting for 0.1% of the ∑₆₃PFAS. The most commonly detected PFCA precursors were 8:2 FTSA, 6:2 FTSA and 10:2 FTUCA, which were found in 35%, 24% and 10% of the samples from Örebro. Of the PFSA precursors, FOSAA, MeFOSAA and perfluorohexane sulfonamide (FHxSA), making up 0.8%, 0.2% and 0.1% of the ∑63PFAS respectively. The most frequently detected PFSA precursors were
MeFOSAA FBSA and FOSAA, found in 24%, 14% and 14% of the samples, respectively. The most commonly detected novel PFAS was PFECHS, which was detected in 22 samples from Örebro. The average PFECHS concentration was 0.016 ng/g, with a maximum of 0.078 ng/g. Another detectable novel PFAS was 6:2 Cl-PFESA; it was found in 4 samples, with a maximum concentration of 0.008 ng/g.
From the 30 samples from Örebro, 22 had EOF levels above the LOD (Figure 5-5-2). Their average EOF concentration was 6.67 ng F/g and the known PFAS (iPFAS) accounted for 25% of it and the remaining 75% was unidentified. The samples with the next highest fraction of UOF were nr. 7, 13 and 24; with 93% of their EOF content remaining unidentified.
30
Figure 5-5-1. PFAS homologue profiles of the samples collected in Örebro; n.a. – not all data was possible to acquire, thus the PFAS homologue profile was not constructed for that sample.
31
Figure 5-5-2. Fluorine mass balance of the samples collected in Örebro; n.a. – not all data was possible to acquire, thus mass balance analysis for that sample was not performed.
32
5.6. Uppsala
The PFAS homologue profiles of the samples from Uppsala were dominated by long-chain PFSAs, on average making up 81% of the ∑63PFAS (see Figure 5-6-1). The long-chain PFSAs
were dominanted by PFOS (L-PFOS - 24% of the ∑63PFAS, 3/4/5-m-PFOS - 22%, 6/2-m-PFOS
- 7.3%), followed by PFHxS (22%), and PFHpS (4.6%); all of these compounds were found in all samples from Uppsala. Both short-chain PFSAs, PFBS and PFPeS made comparatively low contributions to the ∑63PFAS; PFBS accounted for 0.1% of the ∑63PFAS and PFPeS for another
0.6%. PFBS was detected in 31% of the samples and PFPeS was 83%. For the ultra-short chain PFSAs, PFEtS was found in 79% of the samples (maximum concentration of 0.03 ng/g); PFPrS was found in only one sample (0.01 ng/g).
The long-chain PFCAs that made the highest contribution to the ∑63PFAS PFOA (8.6% of
∑63PFAS), PFNA (3.7%), PFDA (1.7%) and PFUnDA (1.7%). The detection frequencies
decreased with increasing perfluorinated carbon chain length. PFOA was found in all samples from Uppsala while PFNA was found in 97% of samples, PFDA and PFUnDA were found in 90% and 52% of the samples, respectively. Of the short-chain PFCAs, only PFPeA was found in one sample at trace levels. The ultra-short chain PFCAs, TFA and PFPrA, were detected in 83% and 7% of the samples respectively.
The PFCA and PFSA precursors made up 0.7% of the ∑63PFAS. The most abundant PFCA
precursors were 10:2 FTUCA and 8:2 FTSA, both accounting for 0.1% of ∑₆₃PFAS. The most commonly detected PFCA precursors were 8:2 FTSA and 6:2 FTSA, which were found in 28% and 17% of the samples from Uppsala; 10:2 FTUCA was detected in one sample. Of the PFSA precursors, FOSAA and MeFOSAA, making up 0.2% and 0.1% of the ∑63PFAS respectively.
The most frequently detected PFSA precursors were FOSA and MeFOSAA, found in 35% and 17% of samples respectively.
The most commonly detected novel PFAS was PFECHS, which was detected in 25 samples from Uppsala. The average PFECHS concentration was 0.024 ng/g, with a maximum of 0.11 ng/g. In addition, both ADONA and 6:2 Cl-PFESA were detected at trace amounts in 4 and 3 samples, respectively.
Fluorine mass balance analysis was performed for 28 samples from Uppsala, of those 26 had EOF levels above the LOD (Figure 5-6-2). Their average EOF concentration was 4.95 ng F/g and the identified PFAS (iPFAS) accounted for 57% and the remaining 43% being of unidentified origin. Samples nr. 13, 14 and 21 were the only samples from Uppsala where all of their EOF was explained by the monitored PFASs. Sample nr. 27 had the highest fraction of UOF – 89%. The samples with the highest EOF content were nr. 17, 7 and 16; their average EOF concentration being 9.94 ng F/g, with 53% of it being UOF.
33
Figure 5-6-1. PFAS homologue profiles of the samples collected in Uppsala; n.a. – not all data was possible to acquire, thus the PFAS homologue profile was not constructed for that sample.
34
Figure 5-6-2. Fluorine mass balance of the samples collected in Uppsala; n.a. – not all data was possible to acquire, thus mass balance analysis for that sample was not performed.
35 6. Discussion
6.1. Legacy PFAS
The main contributors to PFAS exposure, regardless of location, remained the legacy compounds (e.g., PFHxS, PFOS, long-chain PFCAs). This could be attributed to their widespread and historical use. Whilst they have been phased out, their comparatively long half-lives in humans (in the order of several years [54]) result in these compounds still dominating the PFAS burden. The Swedish Food Agencies PFAS 11 (C4, C6 and C8 PFSAs, C4-C10 PFCAs and 6:2 FTSA) were commonly found on average made up 89% of the ∑63PFAS, more
detailed information is provided by Figure 6-1-1.
Figure 6-1-1. Comparison between sum PFAS 11 (C4, C6 and C8 PFSAs, C4-C10 PFCAs and 6:2 FTSA; analytes suggested by the Swedish Food Agency) and sum PFAS 63 (all PFAS monitored in this study).
Figure 6-1-2 shows the average concentrations of C6-C12 PFCAs and C6-C10 PFSAs in the whole blood samples analysed in this study. Of the PFCAs, the C6-C7 compounds were present at low levels, while PFOA (C8 compound) had the highest average concentration (0.56 ng/g). The PFCAs with a longer carbon chain had decreasing concentrations in line with increasing chain-length. The average concetration of PFNA (0.27 ng/g) was approximately 50% of that of PFOA. The levels of PFOA in this study are lower than those reported in 2013 by Bjermo et al.[55], they analysed nearly 300 serum samples from across Sweden and found an average PFOA concentration of 1.1 ng/g (converted for whole blood comparison). However, the levels of PFNA and PFDA were comparable. The results from Bjermo et al. were obtained from serum samples taken in 2010-2011, approximate correction factors were taken from Ehresman et al. [56] to convert their results from serum samples to whole blood basis. A recent study from Colles et al. [57], serum samples from Belgium (collected in 2014), found considerably higher levels for PFOA – 1.4 ng/g (converted for whole blood comparison); these are the results for their reference group aged 50-65.
36
The overall picture was quite different for PFSAs – while the most abundant compound was L-PFOS, a C8 compound, it’s branched isomers (3/4/5-m-PFOS and 6/2-m-PFOS) made a significant contribution as well. Furthermore, C6 and C7 PFSAs (PFHxS and PFHpS) made significant contributions towards the overall PFAS budget as well, while C9 and C10 compounds (perfluorononane sulfonate - PFNS and perfluorodecane sulfonate - PFDS) were not detected. The levels of PFOS in this study are lower than those reported in 2013 by Bjermo et al.[55]; they found an average PFOS concentration of 5.5 ng/g (converted for whole blood comparison) while in this investigation it was 1.6 ng/g. A similar trend was visible for PFHxS, Bjermo et al. reported 1.1 ng/g (converted for whole blood comparison) and this study the average PFHxS concentration was 0.6 ng/g. In the study from Colles et al. the levels of PFOS were higher (3.8 ng/g, converted for whole blood comparison), while PFHxS levels (1.0 ng/g, converted for whole blood comparison) fell between those observed in this study and earlier by Bjermo et al.
Plasma and sera samples were commonly used for different biomonitoring study. However, a recent study showed different preferential binding of some PFASs (FOSA and PFHxA) to wholeblood [51] and thus no detection of these compounds in plasma or sera does not reflect actual human exposure to these compounds. Current investigation demonstrated human exposure to these chemicals and some short-chain compounds directly or indirectly or both, though at lower concentrations, when compared with other long-chain compounds.
Figure 6-1-2. The average concentrations of selected PFAAs in human whole blood.
6.2. Precursor compounds
One possible source for the continuing presence of these legacy compounds (PFAAs) could be exposure to precursor compounds. A previous study has shown that precursor compounds can be metabolized into more stable PFAAs [39]. This hypothesis is supported by the detection of some of the precursor compounds (e.g. 6:2 FTSA, 8:2 FTSA, MeFOSAA) in the whole blood samples, more details are shown in Figure 6-2-1. As these compounds are constantly metabolized, high levels were not expected. However, their metabolites – PFAAs, are known to be both persistent and bioaccumulative[58].
37
Figure 6-2-1. Detection frequencies of selected PFAA precursor compounds.
6.3. Novel PFAS
The most commonly detected novel PFAS was PFECHS, which was detected in more than 80% of the samples, with a mean concentration of 0.02 ng/g and a maximum concentration of 0.11ng/g. The detection frequencies for ADONA and 6:2 Cl-PFESA were far lower, found in 17% and 12% of the samples, respectively. Their maximum concentrations were 0.04 ng/g and 0.02 ng/g correspondingly.
A recent study by Miaz et al. [59] on serum samples from Uppsala found similar levels of PFECHS, ranging from 0.03 ng/g to 0.14 ng/g in whole blood (converted from reported values for serum assuming the conversion factor as other PFAAs). It is known that the chemical has a specific use in aircraft hydraulic fluids; the ubiquitous occurrence of this chemical, even though found at low concentration, suggest other use in our daily life. In their study neither ADONA or 6:2 Cl-PFESA were detected. In a 2017 paper by Fromme et al. [60] they detected trace levels (above 0.1 ng/g in whole blood) of ADONA in samples from Germany, the detection rate was just below 7%. The lower detection rate in the latter study may well be due to higher LOQ levels. Higher levels, when compared to this study, of 6:2 Cl-PFESA have been reported in China (0.9 ng/g by Chen et al. [61] and 3.1 ng/g by Pan et al. [62]); likely due to historical use of the compound in the metal plating industry [63].
6.4. Fluorine mass balance
Fluorine mass balance analysis is a useful tool to estimate the overall levels of organofluorine compounds. In the current study, EOF data from from the combustion ion chromatograph was compared with the levels of individual PFAS from target analysis. A total of 131 samples had EOF levels sufficiently high to be quantified and on average only one third of it could be explained for by the target analysis. As mentioned above, the main drivers of PFAS exposure are still PFAAs, expanding the list of target analytes has not resulted in a significant reduction of the unknown organofluorine fraction.
38
Figure 6-4-1 presents the average EOF level for each location alongside the amount of fluorine explained for by either all 63 analytes included in this study (∑63PFAS) or the 11 analytes
selected for monitoring by the Swedish Food Agency (∑11PFAS). On average, the inclusion of
the additional 52 analytes only explained an additional 3% of the overall EOF.
Figure 6-4-1. Average levels of extractable organofluorine (EOF), fluorine from all analytes included in this study (∑63PFAS) and fluorine from the PFAS11 of the Swedish Food Agency (∑11PFAS) for each location.
The average EOF levels in this study were similar to those observed from Japan (6.9 ng F/g, control group [47] and within the range (<6 – 43.4 ng/mL) found in Chinese population collected from 2004 [48]. The average EOF level of Umeå was between those observed from Japan (6.9 ng F/g) and of samples from USA (24.1 ng F/g) analyzed by Miyake in the aforementioned study.
39 7. Findings
• Swedish people were exposed to 40 different PFAS and unidentified organofluorine compounds
• The average sum concentration of the ∑63PFAS in all whole blood samples was
5.6 ng/g (or 11.2 ng/g after converted for comparison with plasma/sera)
• Long-chain PFSAs (mainly PFOS and PFHxS) and long-chain PFCAs (mainly PFOA) accounted for 77-81% and 16-22% of the ∑63PFAS respectively
• Short-chain PFCAs and PFSAs, ultra-short chain PFSAs, PFCA and PFSA precursors and novel PFAS contributed up to 0.8 % to the ∑63PFAS
• Several novel PFASs (PFECHS, ADONA, 6:2 Cl-PFESA) were detected in the samples
• Several ultra-short chain PFASs (PFPrS, PFEtS, PFPrA, TFA) were also detected in the samples
• The average ∑63PFAS levels in different municipalities are listed as follows:
6.3 ng/g in Uppsala, 6.3 ng/g in Stockholm, 6.0 ng/g in Malmö, 4.7 ng/g in Örebro and 4.7 ng/g in Umeå
• 89 % of the samples showed detectable EOF levels ranging from below LOD to 49.7 ng F/g with an average of 7.8 ng F/g
• The average EOF levels in different municipalities were 13.4 ng F/g in Umeå, 8.31 ng F/g in Stockholm, 6.67 ng F/g in Örebro, 5.40 ng F/g in Malmö and 4.95 ng F/g in Uppsala.
• The fraction EOF explained (indentified PFAS, iPFAS) in the samples are listed as follows: Uppsala (57%), followed by Malmö (52%) , Stockholm (36%), Örebro (25 %), and Umeå (15 %)
8. Conclusions and Future Work
This screening study has shown, that monitoring only a few compounds (e.g. PFOS and PFOA) is insufficient to estimate human exposure to PFASs. The PFOS isomers monitored in this investigation together on average accounted for 36% of the ∑63PFAS. Monitoring program
should also look at PFOS isomers.
While the inclusion of 33 precursor compounds did not significantly improve the fluorine mass balance in humans, the occurrence of several precursor compounds could provide valuable information regarding human exposure routes and contribute to the presence of PFAAs in human samples.
Novel PFASs (PFECHS, ADONA and 6:2 Cl-PFESA) were present in lower concentrations than the legacy PFAS and thus made only a negligible contribution to the fluorine mass balance. However, it may be too soon to disregard monitoring these compounds just yet, studies have indicated the for example 6:2 Cl-PFESA could cause neural toxicity [64]. Since they are used as substitutes for legacy PFAAs, a longitudinal study would be first required. Furthermore, it is known that PFECHS has a specific use in aircraft hydraulic fluids; the ubiquitous occurrence of this chemical, even though found at low concentration, suggest other use in our daily life. Further investigation to its source is warranted.
While most locations in Sweden had comparable extractable organofluorine (EOF) levels with samples from Umeå stood out with somewhat higher EOF content. The EOF levels in current investigation showed an average of 7.8 ng F/g for the general population, which may suggest EOF measurement as a useful tool to detect human exposure to a PFAS source. For example, the samples nr 24-30 from Umeå showed low levels (1.77 – 12.5 ng/g) of ∑63PFAS; however,
40
the EOF ranged 32.5 – 48.7 ng F/g, when compared to other individuals whose EOF levels ranged from below LOD up to 13.9 ng F/g. These individuals may have been exposed to unknown organofluorines that warrant further investigation. Nevertheless, on average the ∑63PFAS explained 37% of the EOF content, just over 60% of the fluorine remained
unexplained. The results showed that humans have exposed to unidentied organofluorine. Current investigation reported ∑63PFAS and EOF levels from 5 municipalities; further
investigation into other municipalities (e.g. Gothenburg and Linköping) provides a more complete picture of human exposure to PFAS and EOF in Sweden.
As the origin, species and thus the potential health effects of the unknown organofluorine (UOF) compounds are not determined, they pose a potential hazard for the public health. This study was able to quantify target PFASs in all samples. For the EOF method, one of the largest challenges is its’ relatively high limit of detection (LOD). One option to circumvent this issue is to use larger sample volume. However, for a more widespread adoption of the EOF and by extention the fluorine mass balance method is dependent on lowering the limits of detection, thus permitting the use of smaller sample volumes.
Additional steps are still required to close the fluorine mass balance, necessitating the use of different methods. As the number of potential PFASs is increasing, more robust analytical tools would be needed. While CIC is one of them, it provides no structural information regarding the origins of the EOF content. One option would be to utilize high-resolution mass spectrometry and methods like suspect screening. This would permit screening for more compounds, but a significant drawback will be the time required for data analysis. Thus, a method such as total oxidizable precursor assay (TOPA) may be a more efficient route to take. In the case of TOPA, precursor compounds are oxidized into more stable PFAAs, that are also easier to measure. By comparing the levels of PFAAs before and after oxidation, it can provide clues about the types of unidentified PFASs present in the sample.
As discussed above, plasma and sera samples were commonly used for different biomonitoring study. A recent study showed different preferential binding of some PFAS (FOSA and PFHxA) to whole blood [51]. Further work should also compare the PFAS, especially those novel PFASs and EOF levels among plasma, sera and wholeblood samples to understand the representation of results from different matrices.
In brief, this study has laid the foundation for further work, which will be required in the following areas:
• Use of high-resolution mass spectrometry to rapidly screen for a wider range of compounds – suspect screening.
• Inclusion of total oxidizable precursor assay to the fluorine mass balance approach in order to convert possible precursor compounds into more readily measurable PFAAs. • Instrumental development in order to further lower the limits of detection of combustion
