Nontarget Time Trend Screening in Human Blood - Environmental

May 25, 2018 - Chiaia-Hernández, A. C.; Günthardt, B. F.; Frey, M. P.; Hollender, J. Unravelling Contaminants in the Anthropocene Using Statistical ...
1 downloads 0 Views 768KB Size
Subscriber access provided by University | of Minnesota Libraries

Ecotoxicology and Human Environmental Health

Non-target time trend screening in human blood Merle M. Plassmann, Stellan Fischer, and Jonathan P. Benskin Environ. Sci. Technol. Lett., Just Accepted Manuscript • DOI: 10.1021/acs.estlett.8b00196 • Publication Date (Web): 25 May 2018 Downloaded from http://pubs.acs.org on May 25, 2018

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 14

Environmental Science & Technology Letters

Non-target Time Trend Screening in Human Blood Merle M. Plassmann,1* Stellan Fischer,2 Jonathan P. Benskin1 1. Department of Environmental Science and Analytical Chemistry (ACES), Stockholm University, SE-106 91 Stockholm, Sweden 2. Swedish Chemicals Agency (KemI), SE-172 67 Stockholm, Sweden *Corresponding author: Merle Plassmann Contaminant Chemistry Unit Dept. of Environmental Science and Analytical Chemistry (ACES) Stockholm University, Stockholm, Sweden Phone: +46 8 674 7159 Fax: +46 8 674 7636 [email protected]

1 ACS Paragon Plus Environment

Environmental Science & Technology Letters

Page 2 of 14

1

Abstract

2

Human biomonitoring (HBM) programs monitor exposure to a limited number of prioritized

3

chemicals resulting in some important substances being overlooked. Non-target analysis shows

4

promise for capturing novel substances, yet the large quantity of data produced by these methods

5

remains challenging. Here we apply a prioritization strategy for temporal non-target HBM data, which

6

shortlists features with increasing time trends, possibly representing substances which are

7

bioaccumulating or to which humans are increasingly exposed. Human whole blood sampled in

8

Germany between 1983 and 2015 was extracted using a modified QuEChERS method, and analyzed

9

by UHPLC-Oribtrap-mass spectrometry. Following alignment, peak detection, grouping and gap

10

filling, up to 14460 features were obtained. This number was reduced to ≤716 using time trend ratios

11

and Spearman’s rank correlation coefficients to identify features which increased over the 32-year

12

time series. Increasing features were investigated further using the KemI market list database (which

13

prioritizes based on human hazard and/or exposure potential) as well as data-dependent product ion

14

scans, followed by MetFrag and mzCloud database searches. Finally, 7 prioritized substances,

15

including 1 pharmaceutical, 2 pesticides and 4 performance chemicals, were confirmed using

16

standards, demonstrating the potential of time trend screening as a data reduction strategy for non-

17

target HBM data.

18

2 ACS Paragon Plus Environment

Page 3 of 14

Environmental Science & Technology Letters

19

Introduction

20

Every year humans are exposed to an increasingly complex mixture of anthropogenic substances,1 via

21

direct contact with consumer products (e.g. personal care products), or through intake of contaminated

22

exposure media (e.g. dust, air, food and drinking water). Human Biomonitoring (HBM) programs

23

evaluate exposure within a population through analysis of biological matrices.2 These programs focus

24

on characterizing exposure to a limited number of priority substances. For example, since 1999, the

25

National Health and Nutrition Examination Survey (NHANES) in the United States has carried out

26

biomonitoring of up to 346 environmental chemicals (or their metabolites) in blood, serum, and urine

27

samples.3 Similar initiatives such as the Canadian Health Measures Survey4 and the HBM4EU

28

biomonitoring initiative5, include even fewer monitored substances.

29 30

Inclusion of a particular target within a biomonitoring program is usually based on existing exposure

31

or hazard data.6 For many emerging contaminants - and indeed transformation products - few hazard

32

or exposure data are available. Moreover, highly specific, targeted analytical approaches rarely exist

33

for emerging substances and their transformation products. As a result, numerous chemicals are

34

overlooked or not included in biomonitoring programs. Fortunately, over the last decade non-target

35

screening has emerged as a promising new technique for detecting many hundreds of chemicals

36

simultaneously in a sample.7-9 Most non-target studies to date have been conducted on water samples10

37

but more recently the approach has been applied to biological samples as well.11-13 Non-target

38

screening offers some clear advantages for HBM, yet the enormous quantity of data produced by such

39

approaches remains a challenge. Data reduction strategies which filter and isolate important features

40

for further identification are vital to the success of these methods. Case/control data designs are the

41

most common and have been used, for example, for identification of biomarkers of exposure and

42

effect in green sea turtles from the Great Barrier Reef,11 as well as novel fluorosurfactants in the blood

43

of firefighters from Australia.13 Recently, we proposed another data reduction strategy involving time-

44

trends, specifically for application in temporal trend biomonitoring studies.14 The approach involves

45

flagging features which show an increasing temporal trend, while removing those with a random,

46

decreasing, valley, or peaking trend. Features displaying an increasing time trend may represent

47

emerging bioaccumulative contaminants which should be prioritized for further investigation. Similar

48

conceptual approaches have been used for identifying transformation products7, 15 and for identifying

49

emerging contaminants in sediment cores,16 but to our knowledge time trend screening in human blood

50

has not yet been carried out.

51 52

The objective of the present study was to apply a non-target time trend data reduction strategy to

53

identify possible bioaccumulative contaminants in human blood. We performed non-target analysis on

54

human whole blood samples from Germany, collected between 1983 – 2015. Features displaying an

55

increasing time trend over the time series were subjected to further identification using a combination 3 ACS Paragon Plus Environment

Environmental Science & Technology Letters

Page 4 of 14

56

of data-dependent MS2 (ddMS2) experiments and database mining. Confirmation of a limited number

57

of prioritized substances was achieved using authentic standards, thereby demonstrating the

58

effectiveness of the workflow.

59 60

Materials and Methods

61

Standards and Reagents

62

Standards of benzotriazole-d4, caffeine-d9, cotinine-d3, diglyme-d6, sucralose-d6, sulfamethoxazole-d4,

63

and triethyl-phosphate-d15 were obtained from Toronto Research Chemicals (Toronto, Canada). 13C8-

64

perfluorooctane sulfonate,

65

obtained from Wellington Labs (Guelph, ON, Canada) while tonalide-d3 was obtained from Dr.

66

Ehrenstorfer (Augsburg, Germany). An internal standard (IS) mixture was prepared using the

67

aforementioned standards at a concentration between 5 – 10 µg/mL.

13

C8-perfluorooctanoic acid and

13

C12-hexabromocyclododecane were

68 69

Sample Collection and Handling

70

Anonymized human whole blood samples from female students aged 21-29 years were acquired from

71

the German Environmental Specimen Bank (Münster, Germany), where they were stored at -150ºC. It

72

is assumed that most contaminants are stable under these conditions, but clearly long-term stability

73

studies are needed to confirm this assumption. A total of 6 samples were randomly selected every 4th

74

year from 1983 to 2015, resulting in 9 different years and 54 samples in total. To test whether pooling

75

affected the performance of our time-trend mining approach (as can occur in targeted time trend

76

studies17), we analyzed samples both individually and as yearly pools. The pooled samples were

77

prepared by combining 2 mL of blood from each of the individuals sampled on the same year.

78

Extractions were carried out using a previously developed method.18 Briefly, after thawing at room

79

temperature, blood samples (4 mL) were spiked with 20 µL IS-mixture and then vortexed. Four

80

stainless steel beads, 4 mL acetonitrile, 1.6 g MgSO4 and 0.4 g NaCl were added, followed by

81

thorough mixing using a bead blender for 2 min at 1500 rpm. After centrifugation (10 min at 3000

82

rpm), 2.5 mL of supernatant were removed and concentrated to 300 µL under a stream of N2.

83 84

QA/QC

85

Instrumental drift was monitored by analysing a 20-fold dilution of the IS-mixture and a quality

86

control (QC) sample (a pooled sample from 2015) repeatedly over the course of the sample sequence.

87

Method detection limits (MDLs) for the 11 ISs were derived from repeated analysis of the QC sample

88

(see SI for details). IS percent recoveries were determined by comparing their peak area in samples to

89

their peak area in a solution prepared in acetonitrile. Finally, each batch of 9 samples was extracted

90

together with a method blank consisting of 1 mL MilliQ water in order to monitor for background

91

contamination.

92 4 ACS Paragon Plus Environment

Page 5 of 14

Environmental Science & Technology Letters

93

Instrumental Analysis

94

Sample extracts were analysed using a Dionex UltiMate 3000 ultra-high performance liquid

95

chromatograph (UHPLC) coupled to a Q Exactive HF hybrid Quadrupole-Orbitrap mass spectrometer

96

(Thermo Scientific), equipped with a heated electrospray ionization source. A detailed description of

97

instrumental conditions can be found in the supporting information (SI). Two different MS

98

experiments were carried out using separate runs in positive and negative mode. First, all samples

99

were run in full scan mode (100 – 1000 Da), referred to herein as the “screening experiment”. After

100

data processing, an inclusion list was created based on features increasing over the time series (see

101

Data Handling section). A second experiment was then performed on selected samples (the

102

“identification experiment”) using full scan (100 – 1000 Da) with ddMS2 fragmentation on masses

103

present in the aforementioned inclusion list. In both experiments, a resolution of 120,000 Full Width at

104

Half Maximum (FWHM) at 200 m/z was used in full scan mode while the resolution of the ddMS2

105

scan was set to 15,000 FWHM at 200 m/z with a stepped normalized collision energy of 30, 70 and

106

120%.

107 108

Data Handling

109

Data from the screening experiment were processed using Compound Discoverer 2.0 (Thermo

110

Scientific) using a workflow involving alignment, unknown compound detection, grouping and gap

111

filling (exact parameters see Table S1, SI). All detected features were exported to Excel where they

112

were split into individual (6 from each year) and pooled samples. A blank subtraction was conducted

113

by deleting all features with peak areas 3 or ρ>0.5 for individual samples; TTR>5 or ρ>0.6 for pooled samples) were combined in an

119

inclusion list to record ddMS2 scans during the identification experiment.

120 121

mzCloud and MetFrag search

122

Data from the identification experiment were processed in Compound Discoverer 2.0 using a

123

workflow which included additional mass list search of an internal suspect list (i.e. features with

124

increasing trends in the screening experiment), and an mzCloud search (parameters see Table S1). All

125

features in the resulting list with recorded MS2 spectra were selected for further processing using

126

MetFrag, which combines database searching and fragmentation prediction for small molecules.19

127 128

Prioritization using KEMI market list

5 ACS Paragon Plus Environment

Environmental Science & Technology Letters

Page 6 of 14

129

Features displaying an increasing trend in the screening experiment were also subjected to further

130

investigation using the KEMI market list (including ~30000 substances, available on the NORMAN

131

Suspect Exchange webpage20), which contains chemicals expected to be on the European market. This

132

database facilitates prioritization of suspects based on a human chronic hazard score (1-9; 9 represents

133

the greatest hazard) and a human exposure score (0-27; 27 represents the highest exposure) which are

134

calculated according to existing hazard data and confidential data about use pattern, tonnage, and

135

hazards in a similar manner as described previously.20 The monoisotopic mass derived from the m/z of

136

features with increasing time trends were matched with the database via an in-house script using a 5

137

ppm tolerance. Out of these matches all halogen containing compounds and those with human

138

exposure score ≥15 and/or a human chronic hazard score of ≥3 were selected for further investigation.

139

These thresholds were found to effectively reduce the number of potential matches while prioritizing

140

substances with higher potential of human hazard and/or exposure.

141

Confirmation using authentic standards

142

Standards were obtained for 11 substances tentatively identified by either the ddMS2 +

143

mzCloud/MetFrag workflow or KEMI market list search. These substances were subjected to

144

additional UHPLC-HRMS analyses, using exact mass, isotopic pattern, retention time and MS2

145

fragmentation for confirmation.

146 147

Results and Discussion

148

QA/QC

149

MDLs ranged from 0.01 – 1.6 ng/mL whole blood (Table S2) which are similar to traditional targeted

150

methods, demonstrating the suitability of the method for biomonitoring of environmental pollutants in

151

blood. IS recoveries ranged from 61.9 – 101.7% for most targets (Figure S1, total range 36.3 –

152

101.7%), demonstrating reasonable accuracy of the method. The exceptions were for sucralose-d6 and

153

cotinine-d3 which displayed lower (albeit reproducible) recoveries of 36.3 and 36.8%, respectively.

154

More importantly, IS recoveries in the sequence QCs (run every 10 samples) displayed an absence of

155

trends over the course of the run for most substances, with the exception of tonalide-d3, which

156

displayed a weak yet significant (p=0.0038, r=-0.79) declining trend, and cotinine-d3 which displayed

157

a positive trend (p=0.0012, r=0.84) (Figure S2). Considering that all samples were randomized

158

throughout the run, the slight instrumental drift observed here is unlikely to have a significant effect

159

on the observed time trends. Therefore, sequence correction, as described in some metabolomics

160

studies,21 was not performed.

161 162

Method precision was also acceptable, with IS RSDs in all samples ranging from 13.4 – 33.5 % in

163

positive mode and 13.1 – 26.1 % in negative mode, except for sucralose-d6 which displayed greater 6 ACS Paragon Plus Environment

Page 7 of 14

Environmental Science & Technology Letters

164

variability (63.8 %) due to low peak intensities. The sequence QC sample displayed much lower

165

variability (6.5 – 28.2 % RSD in positive and 3.2 – 18.1 % in negative mode; 32.4 % for sucralose-d6),

166

indicating that random error from the instrument only accounts for a small part of the overall

167

variability. Overall, method performance was comparable to other non-target studies and suitable for

168

the non-target time trend screening performed here.

169 170

Time trend filtering

171

The screening experiment produced 14460 features in individual samples and 13525 features in pooled

172

samples (sum of features in positive and negative mode; Figure 1). These values reflect removal of

173

isotopes and adduct peaks, but may include substances which ionize by both positive and negative

174

mode. Time trend filtering reduced the number of features to 281 in individual samples (TTR>3 or

175

ρ>0.5) and to 716 in pooled samples (TTR>5 or ρ>0.6). The larger number of features in pooled

176

samples (despite using higher thresholds in pooled samples) might reflect both greater variability in

177

individual samples, resulting in lower TTRs or Spearman’s ρ-values, as well as a higher frequency of

178

low abundance features in pooled versus individual samples. The features with positive trends were

179

subjected to further investigation via a) the identification experiment + mzCloud and MetFrag search,

180

and b) searching the KEMI market list.

181 182

Identification experiment + mzCloud and MetFrag search

183

The identification experiment was carried out using an inclusion list based on features shortlisted from

184

the time trend filtering step (i.e. 281 in individual and 716 in pooled samples). Of these, the

185

identification experiment resulted in MS2 spectra for only 115 of these features in individuals and 134

186

in pooled samples. Other features in the inclusion list were below the set threshold for recording MS2

187

data. Targets for which MS2 spectra were obtained were subjected to database mining using mzCloud

188

and MetFrag.

189 190

A total of 17 features in individuals and 16 features in pooled samples produced matches in MetFrag

191

or mzCloud (mzCloud match score ≥80, Table S3). Of these, 8 substances were identified in both

192

individuals and pools and 1 substance (18-β-Glycyrrhetinic acid) was observed in both positive and

193

negative mode. Notably, 7 pharmaceutical substances were only present in a single sample between

194

2007 and 2015 (individual samples), resulting in a high TTR value, but very low or even negative

195

Spearman’s ρ values. Thus, caution is warranted when applying a TTR to individuals. This

196

phenomena may be controlled either by using pooled samples or by removing samples producing the

197

highest peak in the time series and recalculating TTR and ρ. However, to maximize the number of

198

features included in database mining, this was not carried out in the present work. Overall, a total of

199

17 unique substances were tentatively identified in >1 sample (individual or pooled samples).

200 7 ACS Paragon Plus Environment

Environmental Science & Technology Letters

Page 8 of 14

201

KEMI market list

202

A comparison of the monoisotopic mass of positive time trend features in the screening experiment to

203

the KEMI market list resulted in 287 matches for individuals (209 unmatched / 72 with an average of

204

4 matches/feature) and 833 matches for pools (544 unmatched / 172 with an average of 4.8

205

matches/feature). From this, we prioritized 19 halogen-containing compounds and 31 substances with

206

exposure scores ≥15 and/or hazard scores ≥3 (see Table S4). Overall, 50 unique substances were

207

tentatively identified (individual or pooled samples) of which 3 overlapped with the ddMS2 +

208

mzCloud/MetFrag workflow (desloratidine, oxazepam and diazepam).

209 210

Structural confirmation

211

Of the 11 substances for which standards were obtained, structures were confirmed for 4 per-

212

/polyfluoroalkyl substances (PFASs), 2 pesticides and 1 pharmaceutical. Time trends for these

213

substances are illustrated in Figure 2 and a comparison of MS2 spectra and chromatograms in samples

214

versus standards is provided in Figures S3-S11. Ibuprofen and dibenzepin (identified using MetFrag),

215

as well as quinmerac and 1-octanamine (identified using the KEMI market list) were all confirmed

216

negative (Tables S3+S4). Considering that the KEMI list relies on exact masses of the parent ion,

217

while MetFrag utilizes in silico MS2 data, a lack of experimentally-derived MS2 data probably

218

contributed to the misidentification. Such false positives could be reduced by extending

219

experimentally-derived MS2 databases like mzCloud and Massbank with data for environmental

220

pollutants. Increasing PFAS time trends in human serum in samples from the German specimen bank

221

have been previously reported22, 23 and their detection here serves as a positive control for the entire

222

workflow. The pharmaceutical metabolite triclosan-sulfate has been detected in human serum

223

previously,24 but to the best of our knowledge this is the first report of an increasing time trend for this

224

substance. Hydroxychlorothalonil, a transformation product of the EU-approved fungicide

225

chlorothalonil25 was also confirmed. Previously this substance was tentatively identified in human

226

breast milk samples, but time trends were not reported.26 Finally, haloxyfop is an approved pesticide in

227

several European Countries, including Germany.25 To the best of our knowledge, this is the first time

228

this compound has been detected in human blood.

229 230

Perspectives on the use of non-target time trend screening in HBM

231

The three databases employed here have clear advantages and disadvantages for non-target studies

232

focusing on identification of xenobiotics. While MetFrag searches are time consuming, they can lead

233

to identification of unique substances (e.g. transformation products) which are not present in the

234

KEMI market list or mzCloud (e.g. hydroxychlorothalonil and triclosan-sulfate). While the KEMI

235

market list is the only database in the present study to prioritize substances based on hazard and/or

236

likelihood of human exposure (similar approaches have been described using other databases27, 28), it

237

relies exclusively on exact mass searches based on the parent ion, which can result in false positives. 8 ACS Paragon Plus Environment

Page 9 of 14

Environmental Science & Technology Letters

238

MzCloud was by far the most accessible database in the present work, but this is not surprising given

239

its development by Thermo Scientific, the manufacturer of the Compound Discoverer software used in

240

the present work. However, mzCloud currently contains very few environmental pollutants, thereby

241

severely limiting its utility in the present work. Considering the advantages and disadvantages of each

242

of the databases used here, we recommend a combined approach for future non-target HBM studies.

243 244

Overall, this study demonstrated that a non-target approach, combined with time-trend data reduction

245

and database mining can lead to the detection of novel substances which display increasing

246

concentrations over decades in human blood. While we limited ourselves to confirming substances

247

with positive time trends where standards were readily available, there remain 38 of as-of-yet

248

unidentified features with very high TTR or Spearman’s ρ (TTR≥10 or ρ≥0.7, see Table S5). These

249

features may represent contaminants to which humans are increasingly exposed or which are

250

bioaccumulating in human blood over time, and should be prioritized for further identification.

251 252

Supporting Information

253

The supporting information contains figures and tables about method details, identified substances and

254

confirmed compounds and is available free of charge on the ACS Publications website.

9 ACS Paragon Plus Environment

Environmental Science & Technology Letters

Page 10 of 14

255

References

256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303

1. Persson, L. M.; Breitholtz, M.; Cousins, I. T.; de Wit, C. A.; MacLeod, M.; McLachlan, M. S., Confronting unknown planetary boundary threats from chemical pollution. Environ. Sci. Technol. 2013, 47, 12619-22. 2. Calafat, A. M., The U.S. National Health and Nutrition Examination Survey and human exposure to environmental chemicals. Int. J. Hyg. Environ. Health 2012, 215, 99-101. 3. Fourth National Report on Human Exposure to Environmental Chemicals Updated Tables, March 2018, Volume One; Centers for Disease Control and Prevention - U.S. Department of Health and Human Services. 4. Report on Human Biomonitoring of Environmental Chemicals in Canada Results of the Canadian Health Measures Survey Cycle 1 (2007-2009); Health Canada: 2010. 5. Ganzleben, C.; Antignac, J. P.; Barouki, R.; Castano, A.; Fiddicke, U.; Klanova, J.; Lebret, E.; Olea, N.; Sarigiannis, D.; Schoeters, G. R.; Sepai, O.; Tolonen, H.; KolossaGehring, M., Human biomonitoring as a tool to support chemicals regulation in the European Union. Int. J. Hyg. Environ. Health 2017, 220, 94-97. 6. CDC National Report on Human Exposure to Environmental ChemicalsFrequently asked questions. https://www.cdc.gov/exposurereport/faq.html (Accessed February 15, 2018) 7. Li, Z.; Maier, M. P.; Radke, M., Screening for pharmaceutical transformation products formed in river sediment by combining ultrahigh performance liquid chromatography/high resolution mass spectrometry with a rapid data-processing method. Anal. Chim. Acta 2014, 810, 61-70. 8. Terzic, S.; Ahel, M., Nontarget analysis of polar contaminants in freshwater sediments influenced by pharmaceutical industry using ultra-high-pressure liquid chromatography-quadrupole time-of-flight mass spectrometry. Environ. Pollut. 2011, 159, 557-566. 9. Krauss, M.; Singer, H.; Hollender, J., LC–high resolution MS in environmental analysis: from target screening to the identification of unknowns. Anal. Bioanal. Chem. 2010, 397, 943-951. 10. Kaserzon, S. L.; Heffernan, A. L.; Thompson, K.; Mueller, J. F.; Gomez Ramos, M. J., Rapid screening and identification of chemical hazards in surface and drinking water using high resolution mass spectrometry and a case-control filter. Chemosphere 2017, 182, 656-664. 11. Heffernan, A. L.; Gomez-Ramos, M. M.; Gaus, C.; Vijayasarathy, S.; Bell, I.; Hof, C.; Mueller, J. F.; Gomez-Ramos, M. J., Non-targeted, high resolution mass spectrometry strategy for simultaneous monitoring of xenobiotics and endogenous compounds in green sea turtles on the Great Barrier Reef. Sci. Total Environ. 2017, 599-600, 1251-1262. 12. Mollerup, C. B.; Dalsgaard, P. W.; Mardal, M.; Linnet, K., Targeted and nontargeted drug screening in whole blood by UHPLC-TOF-MS with data-independent acquisition. Drug Test. Anal. 2017, 9, 1052-1061. 13. Rotander, A.; Karrman, A.; Toms, L. M.; Kay, M.; Mueller, J. F.; Gomez Ramos, M. J., Novel fluorinated surfactants tentatively identified in firefighters using liquid chromatography quadrupole time-of-flight tandem mass spectrometry and a case-control approach. Environ. Sci. Technol. 2015, 49, 2434-42. 14. Plassmann, M. M.; Tengstrand, E.; Åberg, K. M.; Benskin, J. P., Non-target time trend screening: a data reduction strategy for detecting emerging contaminants in biological samples. Anal. Bioanal. Chem. 2016, 408, 4203-4208. 15. Su, T.; Deng, H.; Benskin, J. P.; Radke, M., Biodegradation of sulfamethoxazole photo-transformation products in a water/sediment test. Chemosphere 2016, 148, 518-25. 10 ACS Paragon Plus Environment

Page 11 of 14

304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345

Environmental Science & Technology Letters

16. Chiaia-Hernández, A. C.; Günthardt, B. F.; Frey, M. P.; Hollender, J., Unravelling Contaminants in the Anthropocene Using Statistical Analysis of Liquid Chromatography–High-Resolution Mass Spectrometry Nontarget Screening Data Recorded in Lake Sediments. Environ. Sci. Technol. 2017, 51, 12547-12556. 17. Bignert, A.; Eriksson, U.; Nyberg, E.; Miller, A.; Danielsson, S., Consequences of using pooled versus individual samples for designing environmental monitoring sampling strategies. Chemosphere 2014, 94, 177-182. 18. Plassmann, M.; Schmidt, M.; Brack, W.; Krauss, M., Detecting a wide range of environmental contaminants in human blood samples—combining QuEChERS with LC-MS and GC-MS methods. Anal. Bioanal. Chem. 2015, 407, 7047-7054. 19. Ruttkies, C.; Schymanski, E. L.; Wolf, S.; Hollender, J.; Neumann, S., MetFrag relaunched: incorporating strategies beyond in silico fragmentation. J. Cheminf. 2016, 8, 3. 20. Fischer, S.; Schymanski, E., KEMI Market List: Organic chemicals potentially identified on the EU market. http://www.normannetwork.com/sites/default/files/files/suspectListExchange/MarketList_Documentation_25July 2017.docx (Accessed October 30, 2017) 21. Thonusin, C.; IglayReger, H. B.; Soni, T.; Rothberg, A. E.; Burant, C. F.; Evans, C. R., Evaluation of intensity drift correction strategies using MetaboDrift, a normalization tool for multi-batch metabolomics data. J. Chromatogr. A 2017, 1523, 265-274. 22. Schröter-Kermani, C.; Müller, J.; Jürling, H.; Conrad, A.; Schulte, C., Retrospective monitoring of perfluorocarboxylates and perfluorosulfonates in human plasma archived by the German Environmental Specimen Bank. Int. J. Hyg. Environ. Health 2013, 216, 633-640. 23. Yeung, L. W. Y.; Mabury, S. A., Are humans exposed to increasing amounts of unidentified organofluorine? Environ. Chem. 2016, 13, 102-110. 24. Wu, J.-l.; Leung, K.-F.; Tong, S.-F.; Lam, C.-W., Organochlorine isotopic pattern-enhanced detection and quantification of triclosan and its metabolites in human serum by ultra-high-performance liquid chromatography/quadrupole time-of-flight/mass spectrometry. Rapid Commun. Mass Spectrom. 2012, 26, 123-132. 25. European Commission. EU Pesticides database. http://ec.europa.eu/food/plant/pesticides/eu-pesticides-database/ (Accessed February 02, 2018) 26. Baduel, C.; Mueller, J. F.; Tsai, H.; Gomez Ramos, M. J., Development of sample extraction and clean-up strategies for target and non-target analysis of environmental contaminants in biological matrices. J. Chromatogr. A 2015, 1426, 33-47. 27. Newton, S. R.; McMahen, R. L.; Sobus, J. R.; Mansouri, K.; Williams, A. J.; McEachran, A. D.; Strynar, M. J., Suspect screening and non-targeted analysis of drinking water using point-of-use filters. Environ. Pollut. 2018, 234, 297-306. 28. Rager, J. E.; Strynar, M. J.; Liang, S.; McMahen, R. L.; Richard, A. M.; Grulke, C. M.; Wambaugh, J. F.; Isaacs, K. K.; Judson, R.; Williams, A. J.; Sobus, J. R., Linking high resolution mass spectrometry data with exposure and toxicity forecasts to advance highthroughput environmental monitoring. Environ. Int. 2016, 88, 269-280.

11 ACS Paragon Plus Environment

Environmental Science & Technology Letters

346 347 348

Page 12 of 14

Figure 1: Schematic workflow including the number of peaks at each step and confirmed compounds. ‘+’: analysed in positive mode and ‘-‘: analysed in negative mode.

349

12

ACS Paragon Plus Environment

Page 13 of 14

Environmental Science & Technology Letters

350 351 352 353

Figure 2: Trends of confirmed substances detected in individual samples (coloured circles) and pooled samples (black circles except for 8:2 FTS with blue circles), including smoothed trend lines (loess regression) and 90% confidence intervals (coloured for individual and grey for pooled samples)

354

13 ACS Paragon Plus Environment

Environmental Science & Technology Letters

Page 14 of 14

355 356

357 358

14 ACS Paragon Plus Environment