Perfluoroalkyl acids (PFAAs) - ACS Publications

7,8. 58. PFAAs could cross the placenta and are excreted in mother´s milk. .... MINITAB 15® Statistical Software for Windows was used for all statis...
3 downloads 3 Views 577KB Size
Subscriber access provided by Kaohsiung Medical University

Ecotoxicology and Human Environmental Health

Perfluoroalkyl acids (PFAAs) in serum from 2-4-monthold infants – influence of maternal serum concentrations, gestational age, breastfeeding and contaminated drinking water Irina Gyllenhammar, Jonathan P. Benskin, Oskar Sandblom, Urs Berger, Lutz Ahrens, Sanna Lignell, Karin Wiberg, and Anders Glynn Environ. Sci. Technol., Just Accepted Manuscript • DOI: 10.1021/acs.est.8b00770 • Publication Date (Web): 15 May 2018 Downloaded from http://pubs.acs.org on May 16, 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.

1

Perfluoroalkyl acids (PFAAs) in serum from 2-4-month-old infants – influence of

2

maternal serum concentrations, gestational age, breastfeeding and contaminated

3

drinking water

4

Irina Gyllenhammar, † Jonathan P. Benskin,‡ Oskar Sandblom,‡ Urs Berger,§ Lutz Ahrens,#

5

Sanna Lignell, † Karin Wiberg,# Anders Glynn¤ †*

6



7

Uppsala, Sweden

8 9

Department of Risk and Benefit Assessment, National Food Agency, P.O. Box 622, 751 26



Department of Environmental Science and Analytical Chemistry (ACES), Stockholm

University, SE-106 91 Stockholm, Sweden Helmholtz Centre for Environmental Research – UFZ, Department Analytical Chemistry,

10

§

11

Permoserstr. 15, 04318 Leipzig, Germany

12

#

13

Sciences, Box 7050, 750 07, Uppsala, Sweden

14

¤

15

Agricultural Sciences, Box 7050, 750 07, Uppsala, Sweden

Department of Aquatic Sciences and Assessment, Swedish University of Agricultural

Department of Biomedical Sciences and Veterinary Public Health, Swedish University of

16 17 18

*Corresponding author:

19

Anders Glynn

20

Department of Biomedical Sciences and Veterinary Public Health, Swedish University of

21

Agricultural Sciences, Box 7050, 750 07, Uppsala, Sweden

22

Telephone: int + 46 18 672091

23

E-mail: [email protected]

24 1

25 26

Abstract

27

Little is known about factors influencing infant perfluorinated alkyl acid (PFAA)

28

concentrations. Associations between serum PFAA concentrations in 2-4-month-old infants

29

(n=101) and determinants were investigated by multiple linear regression and General Linear

30

Model (GLM) analysis. In exclusively breastfed infants, maternal serum PFAA

31

concentrations 3 weeks after delivery explained 13% (perfluoroundecanoic acid, PFUnDA) to

32

73% (perfluorohexane sulfonate, PFHxS) of infant PFAA concentration variation. Median

33

infant/maternal ratios decreased with increasing PFAA carbon chain length from 2.8 for

34

perfluoroheptanoic acid (PFHpA) and perfluorooctanoic acid (PFOA) to 0.53 for PFUnDA,

35

and from 1.2 to 0.69 for PFHxS and perfluorooctane sulfonate (PFOS). Infant PFOA,

36

perfluorononanoic acid (PFNA) and PFOS increased 0.7-1.2% per day of gestational age.

37

Bottle-fed infants had 2 times lower mean concentrations of PFAAs, and a higher mean

38

percentage of branched (%br) PFOS isomers, than exclusively breastfed infants. PFOA,

39

PFNA and PFHxS increased 8-11% per week of exclusive breastfeeding. Infants living in an

40

area receiving PFAA-contaminated drinking water had 3-fold higher mean perfluorobutane

41

sulfonate (PFBS) and PFHxS concentrations, and higher mean %br PFHxS. Pre- and post-

42

natal PFAA exposure significantly contribute to infant PFAA serum concentrations,

43

depending on PFAA carbon-chain length. Moderately PFBS- and PFHxS-contaminated

44

drinking water is an important indirect exposure source.

45

2

46

TOC Graphic

47 48 49

3

50

51

Introduction

52

Per- and polyfluoroalkyl substances (PFASs) are synthetic fluorinated substances used in

53

numerous commercial applications and occurring ubiquitously in humans and the

54

environment. Sources of human exposure include food, drinking water, dust, air and the use

55

of PFAS-containing consumer products.1-4 Among the ~3000 known PFASs,5 perfluoroalkyl

56

acids (PFAAs) are of greatest concern due to their persistence and chain length-dependent

57

bioaccumulation potential.6 PFAAs may be formed from degradation of other PFASs

58

(commonly referred to as PFAA precursors).7,8

59

PFAAs could cross the placenta and are excreted in mother´s milk.9-11 Modeling of

60

infant exposure to some PFAAs suggests that in utero exposure and breastfeeding are

61

important determinants of blood concentrations.12,13 Food, drinking water, dust and air

62

contribute to exposure when the child gets older.4,14-17 High early life exposure to some

63

PFAAs is related to lower birth weight and immune toxicity,

64

determinants of PFAS concentrations in infant blood is urgently needed for identification of

65

infant exposure sources. Moreover, as pointed out by Loccisano et al. 12 more data on PFAA

66

concentrations in infants is crucial for development/refinement of infant physiologically-

67

based pharmacokinetic (PBPK) models, which are important for risk assessment. To our

68

knowledge only one study has reported PFAA concentrations in infants (6 months old),21

69

showing higher median concentrations of perfluorohexane sulfonate (PFHxS),

70

perfluorooctane sulfonate (PFOS), perfluorooctanoic acid (PFOA) and perfluoronona noic acid

71

(PFNA) in infant serum than in cord serum from the same study participants. The increase

72

was most likely due to exposure during breastfeeding.21

18-20

and knowledge of

4

73

The aim of the present study was to investigate maternal to infant transfer of PFHxS

74

(linear and branched isomers), PFOS (linear and branched isomers), perfluoroheptanoic acid

75

(PFHpA), PFOA, PFNA, perfluorodecanoic acid (PFDA), and perfluoroundecanoic acid

76

(PFUnDA), as well as influence of potential determining factors of infant serum

77

concentrations, such as maternal PFAA concentrations close to delivery, duration of in utero

78

exposure (days of gestational age at delivery), duration of breastfeeding, infant weight gain,

79

and maternal exposure to PFAA-contaminated drinking water.

80

81

Methods

82

Study Population

83

The study is based on samples and data collected within the POPUP cohort (Persistent

84

Organic Pollutants in Uppsala Primiparas), an on-going investigation of POPs in first-time

85

mothers and their children that started in 1996.22 The present study is based on recruitment of

86

mother/child pairs 1996-1999. From early fall 1996 to May 1999, 395 randomly selected

87

primipara women in late pregnancy were asked to participate in the study.23 The Swedish

88

speaking women had been recruited in early pregnancy as controls in a case-control study of

89

caffeine as a risk factor for miscarriage.24 In total 325 women (82%) agreed to participate in

90

the POPUP study.

91

The participating women were asked if a maternal blood sample could be taken 3 weeks

92

after delivery and an infant sample about 3 months after delivery. The mothers were Swedish-

93

born and had singleton births with no birth defects. Almost all women had a full-term

94

pregnancy (37–42 weeks), except in a few cases with the shortest gestational age being 35

95

weeks and the longest 43 weeks. Women in the POPUP cohort are homogenous regarding

96

ethnic background. In 2006, only about 4% of the Swedish population had one or two foreign5

97

born parents.25 Maternal sampling 3 weeks after delivery was accepted by 211 participants.

98

Infant sampling was accepted by 138 participants. Due to limited blood volumes 101 maternal

99

and 107 infant samples were available for PFAA analyses (Table 1). In total 101 matched

100

mother/infant samples were available. Informed consent was obtained from all participating

101

women. The study was approved by the Ethics Committee of the Medical Faculty at Uppsala

102

University (2004:M-177).

103 104

Interviews, Questionnaires and Blood sampling

105

In-person interviews, using a structured questionnaire, were conducted at 6-12 and 32-34

106

completed gestational weeks.24 Height of the women was measured and women were weighed

107

on both interview occasions. Data on maternal characteristics included age, pre-pregnancy

108

body mass index (BMI), years of education, place of residence, and smoking habits. At the

109

infant sampling, the mothers answered a self-administered questionnaire including questions

110

about delivery, and personal characteristics not covered by the interviews, such as data on

111

breast-feeding. Vacutainer® or Vacuette® serum tubes were used for blood sampling, and after

112

separation serum was stored at -20°C until analysis.

113 114

Chemical Analyses

115

The target analytes included C 4 , C6 and C8 perfluoroalkane sulfonic acids (PFSAs; PFBS,

116

PFHxS, PFOS) and C 6 -C15 perfluoroalkyl carboxylic acids (PFCAs; PFHxA, PFHpA, PFOA,

117

PFNA, PFDA, PFUnDA, PFDoDA, PFTrDA, PFTeDA, PFPeDA; for details see Supporting

118

Information, Table A1). The serum samples were analyzed as described previously.26 In short,

119

serum (0.1 g for infants and 0.5 g for adults) was spiked with internal standards (Supporting

120

Information, Table A2) and extracted with acetonitrile (Honeywell CHROMASOLV™,

121

≥99.95% purity) in an ultrasonic bath. The concentrated extract underwent dispersive clean-up

6

122

with graphitized carbon (Supelclean™ ENVI-Carb™, Sigma-Aldrich). Aqueous ammonium

123

acetate (2mM; EMSURE®, Sigma-Aldrich) and volumetric standards (13 C8 -PFOS and 13 C8 -

124

PFOA; both from Wellington Labs) were added before instrumental analysis on an Acquity

125

ultra performance liquid chromatography system (UPLC) coupled to a Xevo TQ-S tandem

126

mass spectrometer (both Waters Corp., Milford, MA, U.S.) operated in negative electrospray

127

ionization, multiple reaction monitoring mode. The mobile phase consisted of water and

128

acetonitrile (ACN; both containing 2mM ammonium acetate), with the relative proportions

129

varied under gradient conditions. ACN-based mobile phases such as that used here are highly

130

effective at removing bile acid interferences which may affect PFOS and PFHxS analysis. 27-29

131

Further, instrumental parameters are provided in Supporting Information (Table A2).

132

Quantification was performed by isotope dilution using a 5-point external calibration curve

133

(linear, 1/x weighting), which was run before and after samples. For most targets, analogous

134

isotopically labelled internal standards were available. For PFBS, PFTrDA, PFTeDA, and

135

PFPeDA, a structurally similar internal standard was used (Supporting Information, Table

136

A2). For PFHxS and PFOS, the linear isomer and the sum of all branched isomers (as one

137

signal) were quantified separately using a linear isomer calibration curve. For PFOS,

138

concentrations obtained from the m/z 499/80 and 499/99 ions were averaged, which provides

139

a reasonable estimate of the weight % branched (%br) content in the absence of isomer-

140

specific quantification.30 PFHxS was also monitored using m/z 399/80 and 399/99 ions;

141

however, since the branched isomer peak in the latter transition was often below method

142

detection limits (MDLs), concentrations were only reported from the m/z 399/80 transition.

143

Both a procedural blank (0.1 or 0.5 mL Milli-Q water) and QC sample (pooled human

144

serum analyzed repeatedly in-house) were included with every batch of samples (see

145

Supporting Information, Table A3, for QC performance metrics). In addition, concentrations

146

determined in n = 3 replicates of NIST SRM 1957 were compared to reference values to

7

147

assess the accuracy of the method (results provided in Supporting Information Tables A4 and

148

A5). Measured concentrations in SRM 1957 were consistent with reference values for all

149

targets, while coefficients of variation (CVs) in in-house reference serum ranged from 12 -

150

36% (median 19%; n = 20; Table A3). For targets observable in method blanks, the MDL was

151

based on the mean blank + 3× the standard deviation of the blanks. For those targets with no

152

observable blank contamination, detection limits were based on the concentration producing a

153

signal-to-noise ratio of 3 in sample chromatograms. Analyses of PFAAs in infants and their

154

mothers were accomplished within three different analytical batches in 2013, 2014, and 2016.

155

Batch-specific MDLs are provided in Table A6 of the Supporting Information. Further

156

method validation parameters are provided in Glynn et al. 2012.26

157 158

Statistical Analyses

159

MINITAB 15® Statistical Software for Windows was used for all statistical analyses. When

160

PFAA concentrations were below the method detection limit (MDL), MDL/√2 was taken as

161

an estimated value in the statistical analyses. The PFAA concentrations in maternal serum

162

were measured in different analytical batches and for PFHpA and PFBS the MDL differed

163

between the batches (Supporting Information Table A6). In the statistical analysis including

164

maternal PFAA data, all PFHpA and PFBS data below MDL in the batch with the highest

165

MDL was omitted (batch 3, n=17). The proportions of branched and linear isomers for PFHxS

166

and PFOS were expressed as a percentage of the total concentration. Correlations between

167

infant and mother serum PFAA concentrations were investigated using the Pearson´s

168

correlation test, and only pairs with data >MDL for both the mother and the infant were

169

included in the analyses.

170

For PFAAs with over 75% of infant data >MDL multiple linear regressions were used

171

to analyze if infant serum concentrations are influenced by maternal PFAA concentrations 3

8

172

weeks after delivery (ng g-1 ), gestational age at delivery (days), infant weight gain (% of birth

173

weight) and duration of breastfeeding (weeks) (Table 1). The inclusion of the potential

174

determinants were based on the assumption that they may influence fetal/infant exposure

175

(maternal PFAA concentrations, gestational age, duration of breastfeeding) or infant PFAA

176

distribution volume (infant weight gain). In univariate analyses infant PFAA concentrations

177

did not differ between sexes (Mann-Whitney U test, p>0.05), and infant sex was not included

178

as a covariate in the regression models. In this analysis, only data from infants that were

179

exclusively breastfed during the whole study period from birth to infant blood sampling were

180

included. Stepwise regression analysis was used to estimate how much of the variation in

181

PFAA concentrations that could be explained by the variation of each individual determinant

182

(partial coefficient of determination, R2 ). Criteria for determinant inclusion in the regression

183

model was α50% of concentrations below MDL for PFDoDA, PFTrDA,

224

PFTeDA and PFPeDA (Table 2). The observed concentrations differ from those reported for

225

6-month-old infants from Germany,21 the only previously published infant study we could

226

find in the literature. Different sampling periods (present study:1996-99, German study:2007-

227

2009) and ages of the infants (present study:2-4 months, German study: 6 months) are most

228

likely important determinants of the observed differences.

229

To our knowledge, no other study has reported isomer-specific data on PFHxS and

230

PFOS for infants. We found similar %br PFOS and PFHxS in mothers and infants (Table 2),

231

suggesting that infant isomer patterns to a large extent are dependent on maternal patterns. In

232

Canadian mothers, median %br PFOS (36%) was similar to that of Swedish mothers,

233

although Canadian mothers were sampled almost 10 years after the Swedish mothers. 32 In

234

maternal serum from China, sampled 2015-2016, branched PFHxS accounted for 14%

235

(median) of the total PFHxS and branched PFOS 17% of total PFOS. 33 A comparison of

236

maternal %br PFHxS and PFOS between studies is hampered by differences in how branched

237

content was determined in each study (e.g. isomer-specific versus sum branched isomer

238

quantification using linear and/or individual branched isomer standards). Nevertheless,

239

differences due to ongoing maternal exposure to for instance PFHxS and PFOS precursors in

240

China are likely since perfluorooctane sulfonyl fluoride-based chemistry is still used in

241

China.33

242

Maternal serum/plasma concentrations of PFHxS, PFOS, PFOA, PFNA, PFDA, and

243

PFUnDA during pregnancy and a few weeks after delivery both reflect placental and mother´s

244

milk PFAA transfer, as shown by strong correlations between PFAA concentrations in

245

maternal and cord blood, and between concentrations in maternal blood and mother´s

246

milk.9,10,26,34,35 We found positive correlations between maternal and infant PFAA

11

247

concentrations, with weakening strength as chain length increased of both long-chain PFCAs

248

and PFSAs (Table 3). This suggests that factors other than maternal concentrations become

249

more important as determinants for infant PFAA concentrations as perfluoroa lkyl chain length

250

increases.

251

This is further supported by the decreased strength of association between maternal and

252

infant concentrations with increased PFAA carbon chain length among exclusively breastfed

253

infants, as determined in multiple log-linear regression analyses (Table 4). Variation in

254

maternal PFOA concentrations explained 53% of the infant concentration variation, whereas

255

only 13% of the variation in infant PFUnDA was explained by maternal variation. For

256

PFHxS, 73% of the infant variation was explained by maternal variation, and the

257

corresponding figure for PFOS was 36%. The short-chain PFHpA deviated from this pattern

258

of chain-length dependency, showing a weaker infant/mother correlation than PFOA did

259

(Table 3) and a weaker strength of association (Table 4). For PFAA homologues with less

260

strong infant/mother correlations and associations, contribution of infant exposure sources not

261

studied by us may be important, including ingestion of dust, suction on materials containing

262

PFASs, inhalation of fine particles/air, and skin exposure through use of personal care

263

products.4 To what degree these different exposure pathways contribute to total infant

264

exposure has not been fully characterized, but the contribution of each component most

265

probably varies with infant age.4

266

Our results suggest that PFAA transfer efficiency from mother to infant decreases with

267

increasing perfluoroalkyl chain length, as indicated by the infant/maternal serum

268

concentration ratios (Table 3). On average, PFHpA and PFOA showed higher concentrations

269

in infants than in mothers, PFHxS concentrations were similar, and concentrations for PFNA,

270

PFOS, PFDA and PFUnDA were lower in infants. This is in agreement with the previously

271

observed chain length-dependent decrease in transfer efficiency over the placenta and from

12

272

maternal blood to mother´s milk.9,32,36,37 It has been proposed that decreases in PFAA

273

placental transfer with increased chain length may be due to decreases in water solubility and

274

increases in PFAA affinity to serum albumin.4,37 It has also been hypothesized that there are

275

chain length differences in active PFAA transport in the placenta,4,11 although this hypothesis

276

has to our knowledge not been tested. Similarly, it is not known to what degree mammary

277

gland active PFAA transport is involved in the carbon chain-dependency of PFAA transfer to

278

mother´s milk.

279

In contrast to Fromme et al.21 , our study does not include matched cord blood data. We

280

have previously published cord whole blood data for PFOS, PFOA and PFNA in the Uppsala

281

study 1996-99.26 Assuming a distribution factor of 2 between serum and whole blood, average

282

cord serum concentrations were 2-3 times lower than those in maternal serum sampled in late

283

pregnancy,26 which is in accordance with data from Fromme et al. 21 A comparison of

284

published cord/maternal serum or plasma PFAA ratios with the infant/maternal serum PFAA

285

ratios among Uppsala infants (Table 3) and German infants (plasma) show that the

286

infant/maternal ratios generally are higher than the cord/maternal rations.21,32,36 This strongly

287

suggests that breastfeeding gives a significant contribution to the PFAA exposure of the

288

infants.

289

The similar median ratios of infant/maternal linear and branched PFHxS, as well as of

290

PFOS isomers (Table 3), show that linear and branched isomers were retained in the infants to

291

a similar degree as in their mothers. A few studies have reported less efficient placental

292

transfer of linear PFOS (relative to branched), but found no isomer specific differences in

293

transfer via mother´s milk.10,32,35 To our knowledge, only one study has reported placental

294

transfer efficiencies of PFHxS isomers showing, contrary to PFOS, a more efficient transfer

295

of the linear isomer than of the branched isomers.33 Our results suggest that the more efficient

296

trans-placental transfer of branched PFOS, and less efficient transfer of branched PFHxS, are

13

297

“compensated for” after birth, presumably from mother´s milk, resulting in similar %br

298

isomers in mothers and infants.

299 300

Impact of Duration of Breastfeeding, Gestational Age and Infant Weight Gain

301

Concentrations of PFOA, PFNA, PFDA, PFUnDA, PFHxS and PFOS were lower in

302

exclusively bottle-fed infants than in exclusively breastfed infants, strongly suggesting that

303

mother´s milk is an important source of infant exposure to all these PFAAs (Fig. 1). The

304

results are however uncertain since we only had 5 exclusively bottle-fed infants, but the

305

importance of breastfeeding as an exposure source of all these PFAAs is supported by our

306

findings of lower serum PFAA concentrations also among mixed breast-/bottle-fed infants

307

(Fig. 1). The lower PFAA concentrations in bottle-fed infants are most probably a net result of

308

infant excretion and growth dilution of PFAAs accumulated in utero, and much lower PFAA

309

exposures from infant formula than from mother´s milk.38,39 No exclusively bottle-fed infants

310

were living in a district receiving contaminated drinking water, so it was not possible to

311

determine the contribution of this potential exposure route.

312

The higher %br PFOS in the few exclusively bottle-fed infants than in exclusively

313

breastfed infants (Fig. 1) is most probably a result of the higher placental transfer of branched

314

PFOS isomers than of the linear isomer, and a “mother´s milk compensation” with the linear

315

isomer among exclusively breastfed infants.10,32,35 Branched PFOS isomers are preferentially

316

excreted in urine in adults compared with the linear isomer, 40,41 but excretion of different

317

PFOS isomers in infants apparently was too slow to cause preferential retention of the linear

318

isomer in bottle-fed infants. The reported lower trans-placental transfer efficiency of branched

319

PFHxS isomers than of the linear isomer33 did not result in lower %br PFHxS concentration

320

in the exclusively bottle-fed infants (Fig. 1). Assuming that the excretion of PFHxS was

14

321

negligible during infancy, the results suggests that the transfer of branched PFHxS isomers to

322

mother´s milk is not higher than transfer of the linear isomer.

323

In the multiple regression analyses of PFAA concentrations in exclusively breastfed

324

infants, duration of breastfeeding explained 90% PFAA-contaminated

384

drinking water between 1996 and 1999, than among infants living in the districts not receiving

385

contaminated water (Fig. 2).

386

Similarly as among Uppsala mothers,31 no influence of contaminated drinking water on

387

infant PFOS concentrations was observed (Supporting Information, Table A7), although

388

PFOS concentrations in the contaminated drinking water were only slightly lower than those

389

of PFHxS and much higher than PFBS concentrations. Taken together the results suggest that,

390

in contrast to PFBS and PFHxS, maternal PFOS exposure sources other than drinking water

391

were more important indirect determinants of infant PFAA concentrations.

392

%br PFHxS was significantly higher in infants living in Uppsala districts receiving 10-

393

90% and >90% contaminated water than in infants living in the district receiving

394

uncontaminated water (Fig. 2). We have previously proposed that increased %br PFHxS in

395

adult serum, compared to background in a population, may be an indicator for drinking water

17

396

exposure to PFHxS originating from fire-fighting foams.31 Our infant results suggest that

397

enrichment of branched PFHxS persists also in infants of mothers exposed to drinking water

398

PFHxS.

399 400

Strengths and Limitations

401

Our study includes data on PFBS, PFHxA, PFHpA, PFUnDA, PFDoDA, PFTrDA, PFTeDA

402

and PFPeDA which were lacking in Fromme et al. (2011).21 For many of these PFAA,

403

however, concentrations were generally below LOQ, making it impossible to analyze

404

associations with potential determinants. We have data on branched and linear isomers of

405

PFHxS and PFOS in infants, thus enabling a first look at possible isomer-specific differences

406

in mother/infant dynamics of PFHxS and PFOS. We only had PFAA concentration data from

407

5 infants that were exclusively bottle-fed, making the comparisons with PFAA concentrations

408

in exclusively breastfed infants uncertain. However, with supporting data from partially

409

bottle-fed infants we can nevertheless conclude that breast-feeding is a significant source of

410

infant exposure to PFHxS, PFOS, PFOA, PFNA, PFDA and PFUnDA. Moreover our study

411

includes data on several potential determinants not studied before, such as gestational age

412

(duration of in utero exposure), infant body weight gain (growth dilution), and home address

413

in drinking water districts with differences in drinking water PFAA contamination (indirect

414

drinking water exposure). A weakness is that we did not have data on PFAA concentrations in

415

drinking water ingested by the mothers and maternal drinking water consumption, which

416

would have improved the analyses of contribution of maternal PFAA drinking water exposure

417

to infant serum PFAA concentrations.

418

The Swedish infants were only sampled once after birth and more extensive serial

419

sampling of neonates and infants will give better information about factors influencing

420

serum/plasma PFAA concentrations during early childhood. As shown by the decrease in

18

421

strength of determination with increasing PFAA chain length, there are unknown

422

determinants, not studied by us, that have an important influence on infant PFAA

423

concentration.

19

424

Figures

425

PFOA

0.15

0.5

PFHxS

0.10

re as tfe d

20

*

*

0.06 0.04

ng/g serum

15

ng/g serum

2

1

10 5

0.02

%branched PFHxS

re a B

/b re a

st fe d

st fe d

ed B

ot tle -

B

B

re a

st fe d

st fe d /b re a ot tle B

B

ot tle -

B

B

re a

ot tle f

st fe d

st fe d /b re a

ot tle f B

0 ed

0 ed

0.00

ot tle f

ng/g serum

B

PFOS

3

*

0.08

%branched PFOS 60

%branched isomers

10 8 6 4 2

20

B

re a

st fe d

st fe d /b re a ot tle -

re a B

ot tle f B

ot tle B

st fe d

st fe d /b re a

ot tle f B

ed

0 ed

0

*

40

B

%branched isomers

st fe d B

ot tle B

PFUnDA

/b re a

/b re as tfe d

B

re as tfe d

B

ot tle -

B

/b re as tfe d

0.00

ot tle -

0.0

0.05

ot tle fe d

0

ot tle fe d

0.1

ot tle fe d

2

B

ng/g serum

0.2

0.10

re as tfe d

4

0.3

B

ng/g serum

6

*

*

0.4

B

*

8

ng/g serum

PFDA

PFNA

10

426 427

Figure 1. Concentrations (ng g-1 serum) of selected PFAAs in serum from 2-4-month-old infants

428

sampled 1996-1999 with different degrees of breastfeeding during the study period. ‘Bottle-fed’ (N=5)

429

represents infants which were not breastfed at all prior to sampling. ‘Bottle-/breastfed’ (N=11)

430

corresponds to infants that were exclusively breastfed followed by partial/exclusive bottle-feeding

431

before sampling. ‘Breastfed’ (N=83) represents infants exclusively breastfed from birth until

432

sampling. PFAA concentrations