Chemosphere 118 (2015) 170–177

Contents lists available at ScienceDirect

Chemosphere journal homepage: www.elsevier.com/locate/chemosphere

The effect of ongoing blood loss on human serum concentrations of perfluorinated acids M. Lorber a,⇑, G.E. Eaglesham b, P. Hobson c, L.-M.L. Toms d, J.F. Mueller e, J.S. Thompson e a

Office of Research and Development, United States Environmental Protection Agency, 1200 Pennsylvania Ave, NW, Washington, DC 20460, United States Queensland Health and Forensic Scientific Services, Special Services Organics Group, 39 Kessels Rd., Coopers Plains, QLD 4108, Australia c Sullivan Niccolaides Pathology, PO Box 344, Indooroopilly, QLD 4068, Australia d School of Clinical Sciences and Institute of Health and Biomedical Innovation, Queensland University of Technology, Gardens Point, Brisbane, QLD 4001, Australia e The University of Queensland, National Research Center for Environmental Toxicology (Entox), 39 Kessels Rd., Coopers Plains, QLD 4108, Australia b

h i g h l i g h t s  We investigate if blood loss reduces body burdens of perfluorinated acids, PFAAs.  PFAAs were lower in males requiring regular blood withdrawals for a medical condition.  Pharmacokinetic (PK) modeling was able to duplicate this finding.  PK modeling also showed that menstruation would also reduce PFAAs in women.

a r t i c l e

i n f o

Article history: Received 1 April 2014 Received in revised form 22 July 2014 Accepted 29 July 2014

Handling Editor: Andreas Sjodin Keywords: PFOS PFOA Menstruation Elimination Blood loss PK model

a b s t r a c t Perfluorinated alkyl acids (PFAAs) have been detected in serum at low concentrations in background populations. Higher concentrations haven been observed in adult males compared to females, with a possible explanation that menstruation offers females an additional elimination route. In this study, we examined the significance of blood loss as an elimination route of PFAAs. Pooled serum samples were collected from individuals undergoing a medical procedure involving ongoing blood withdrawal called venesection. Concentrations from male venesection patients were approximately 40% lower than males in the general population for perfluorohexane sulfonate (PFHxS), perfluorooctane sulfonate (PFOS) and perfluorooctanoic acid (PFOA). A simple pharmacokinetic model was used to test the hypothesis that blood loss could explain why adult males have higher concentrations of PFAAs than females, and why males undergoing venesections had lower concentrations compared to males in the general population. The model application generally supported these hypotheses showing that venesection might reduce blood serum concentrations by 37% (PFOA) and 53% (PFOS) compared to the observed difference of 44% and 37%. Menstruation was modeled to show a 22% reduction in PFOA serum concentrations compared to a 24% difference in concentrations between males and females in the background population. Uncertainties in the modeling and the data are identified and discussed. Published by Elsevier Ltd.

1. Introduction Abbreviations: AFFF, aqueous film fire-fighting foams; C, concentration of a PFAA in serum; D, daily intake dose; kP, first order elimination rate; NHANES, National Health and Nutritional Evaluation Survey; PFAA, perfluorinated alkyl acid; PFBA, perfluorobutanoic acid; PFBS, perfluorobutane sulfonate; PFDoDA, perfluorododecanoic acid; PFHxS, perfluorohexane sulfonate; PFNA, perfluorononanoic acid; PFOA, perfluorooctanoic acid; PFOS, perfluorooctane sulfonate; PFPeA, perfluoropentanoic acid; PFUnDA, perfluoroundecanoic acid; PK, pharmacokinetic; Vd, volume of distribution. ⇑ Corresponding author. E-mail address: [email protected] (M. Lorber). http://dx.doi.org/10.1016/j.chemosphere.2014.07.093 0045-6535/Published by Elsevier Ltd.

Perfluorinated alkyl carboxylic and sulfonic acids are two groups of perfluorinated alkyl acids (PFAAs) compounds classes of man-made compounds in use since the 1950’s in various industrial and consumer applications. These include use in oil and water resistant surface coatings, in aqueous film fire-fighting foams (AFFFs), and in the process of fluoropolymer manufacturing (Prevedouros et al., 2006; OECD, 2002). Due to environmental

M. Lorber et al. / Chemosphere 118 (2015) 170–177

concerns regarding their persistence and potential adverse health effects, in recent years their production and use has been limited and discouraged in many areas (NICNAS, 2007, 2008; Wang et al., 2009). The eight carbon anion of the sulfonic acid, perfluorooctane sulfonate, PFOS, has been added to the Stockholm convention which aims to reduce persistent organic pollutants in the environment (Stockholm Convention, 2010). However, PFAAs are still used in the semi-conductor and metal plating industries, and other PFAA containing products, such as specialty AFFFs, treated carpets and textiles, are likely present in stockpiles and inventories until their use or disposal and replacement. PFAAs have been measured in wildlife and human populations from around the globe (Houde et al., 2006). The route of human exposure to these compounds is not entirely elucidated, but a considerable portion appears to come from diet (Fromme et al., 2009; Lorber and Egeghy, 2011; Egeghy and Lorber, 2011). Multiple studies have found significant differences between adult male and female serum concentrations of both PFOS and/ or PFOA (Olsen et al., 2005; Calafat et al., 2007; Toms et al., 2009), with males having higher concentrations than females. For instance in data from South East Queensland, Australia, where the current study was conducted, samples pooled by age (30 individuals per pool) showed higher concentrations of PFOS in all male pools over 16 years by 2.5–5.6 ng mL1. For PFOA the differences were smaller, 0.9–2.4 ng mL1 (Toms et al., 2009). In data from the 2003/2004 National Health and Nutritional Evaluation Survey (NHANES) in the US, the male geometric mean PFOS concentration was 23.3 ng mL1, compared with 18.4 ng mL1 in females. For PFOA the same comparison was 4.5 ng mL1 and 3.5 ng mL1 (results for individuals > 12 years; Calafat et al., 2007). Toms et al. (2014) looked at trends in PFAA serum concentrations of Australians between 2002 and 2011. They found that males had higher concentrations of PFOS, PFOA, PFNA, and PFHxS from about age 15 until age 60, and after age 60, males and females had similar concentrations. This is not consistent in every study, and some have found no significant differences (Kubwabo et al., 2004). However, when observed, the differences have typically been between males and females from the approximate ages of 13–50 years, with little gender differences outside this range (Harada et al., 2005; Ingelido et al., 2010; Toms et al., 2009, 2014). Given this age dependence, it has been suggested that menstruation is an additional route of elimination available to reproductively mature females (Harada et al., 2005), as it is for other xenobiotics (Silvaggio and Mattison, 1994; Soldin and Mattison, 2009). There are certainly other factors affecting the observed patterns in body burden, such as physiological differences, differences in exposure pathways, and the history of child birth in the females studied. Despite this, menstruation and blood loss remains an obvious point of contrast and an intuitive candidate for an additional elimination mechanism. With this as background, we sought to investigate the effect of blood loss on serum PFAA concentrations in humans. This was done via chemical analysis of serum, combined with the use of a simple pharmacokinetic (PK) model. The serum data originated from persons undergoing regular venesection, also termed phlebotomy. Venesection is a medical practice whereby a quantity of blood is withdrawn from an individual in order to relieve symptoms of disorders such as haemochromatosis. We observed that both males and females showed lower PFAA concentrations than their counterparts in the general population (details below). Then, through application of the PK model, we examine whether or not the observed lower male PFAA body burdens can be explained by loss of PFAAs via loss of blood during these procedures. We then additionally use the model to examine whether difference in males and females in the general population can be explained by menstrual loss of blood.

171

2. Materials methods 2.1. Sample collection/pooling Archived de-identified serum samples stored at 20 °C were obtained through Sullivan-Nicolaides Pathology from patients undergoing venesection treatment, for the conditions of either haemochromatosis (99% of donors) or polycythaemia (1% of donors). All samples were originally collected from individuals in the course of unrelated blood work (e.g. health checkups) in 2009. Patient records were available providing only age, date of sample collection (again all collected in 2009), number and dates of venesections since their treatment began in 2004, and the condition requiring venesection. In total, archived serum from 151 individuals was obtained. Composite samples were then created, with each comprised of material from six individual samples, pooled according to: gender, age (three strata: ‘>60 years’, ‘60 years or less’, ‘all ages’), number of venesections since 2004 (two strata: ‘>10’, ‘10 or less’) and time since last venesection before 2009 sampling (two strata: ‘>365 d’, ‘365 d or less’) termed hereafter ‘replenishment time’. We established these strata based on an examination of the data, believing they would reasonably explore the effects of the total volume of serum lost and the replenishment of the body with PFAAs through background intakes. Further, the pooling of samples was an essential component of the study’s ethical requirements. The pooling resulted in a total of 33 composite samples, 23 male and 10 female. We focus here on 15 of the male and 7 of the female composites. All of these 22 were in the stratified age-specific pools: either the ‘>60 years’ or the ‘60 years or less’ groups, depending on age. There were six independent serum samples contributing to each composite, and so these 22 composites represented 132 individuals. The other 11 pools were termed, ‘all ages’ pools. There were 8 male and 3 female ‘all ages’ pools. These ‘all ages’ pools were comprised of serum samples, some of which had also been used in the age-specific pools. Since there was some duplication (repeat usage of individual serum samples) in these ‘all ages’ pools, including them in statistical analyses would be invalid. For this reason, we do not provide and analyze the ‘all ages’ results here; a summary of results from them are provided in Supplementary Material. An age bias is seen in the individuals sampled; specifically that there were a larger number of older individuals in the full cohort of 151 original individuals as compared to younger individuals. This is because the conditions requiring venesection tend to not manifest negative effects until later in life. In the final pooled samples, 9 of the 15 age-specific pools for males and 5 of the 7 age-specific pools for females were for the age range >60 years. Regarding the other pool characteristics, 10 pools were characterized as having greater than 10 venesections, leaving 12 for 10 or less venesections, and 10 pools were greater than 365 d since last venesection and sampling leaving 12 with 365 d or less. To provide a general population ‘control’ group for comparison, data were compared with results from pooled serum samples collected from individuals living in South East Queensland (Toms et al., 2014). To ensure a valid comparison of the two data sets, four of these pools were retrieved and analyzed in duplicate at our laboratory along with the new serum pools. Results of this validation are provided in Supplementary Material. Ethics approval for this study was granted by The University of Queensland Medical Research Ethics Committee.

2.2. Extraction and analysis A solution of isotope labeled PFAAs (1 ng) was added to 1 mL of each pool, consisting of 13C4-perfluorobutanoic acid (13C-PFBA),

172

M. Lorber et al. / Chemosphere 118 (2015) 170–177

13

C2-perfluorohexanoic acid (13C-PFHxA), 13C4-perfluoroocotanoic acid (13C-PFOA), 13C5-perfluorononanoic acid (13C-PFNA), 13C2-perfluorodecanoic acid (13C-PFDA), 13C2-perfluoroundecanoic acid (13C-PFUnDA), 13C2-perfluorododecanoic acid (13C-PFDoDA), 18 O2-perfluorohexane sulfonate (18O-PFHxS), and 13C4-perfluorooctane sulfonate (13C-PFOS). The corresponding native PFAAs (abbreviated in text as for isotope labeled analogs without the 13 18 C/ O-prefix) were quantified using these internal standards and a calibration series consisting of native compounds at varying concentrations (0.1–50 ng/mL) with constant concentrations of mass labeled standards added. Some analytes lacked a corresponding isotope labeled standard. For quantification of perfluoropentanoic acid (PFPeA) the 13C-PFBA internal standard was used, for perfluoroheptanoic acid (PFHpA) 13C-PFHxA was used, and for perfluorobutane sulfonate (PFBS) the 18O-PFHxS internal standard. Limits of detection ranged from 0.07 ng/mL for PFDA to 0.5 ng/mL for PFNA. The extraction method used was a slight modification of the method of Yeung et al. (2009). Briefly, the method involved the addition of 5 mL acetonitrile and shaking on an orbital shaker (150 rpm, 10 min) for protein precipitation and PFAA extraction. This was repeated three times successively, with centrifugation (3500 rpm, 5 min) used between each solvent addition to isolate the extracts. The combined extracts were concentrated and diluted with water before purification via weak anion exchange SPE (WAX–SPE 6 cc 5 mg OASIS, Waters, Milford USA). The only adjustments made to the published method was the use of serum rather than whole blood, concentration of the combined extracts down to a volume of 2.5 mL as opposed to 0.5 mL, and dilution with milli-Q water up to 50 mL as opposed to 10 mL. The finished extracts were analyzed on a HPLC/MS/MS system consisting of a Shimadzu prominence HPLC (Shimadzu corp. Kyoto, Japan) coupled to a QTrap 4000 quadrupole mass spectrometer (AB/MDS Sciex. Concord, Ont. Canada) using negative ESI ionization. The mass spectrometer was operated in scheduled MRM mode, monitoring two transitions for all but PFBA and PFPeA. Signals for several PFAAs were intrinsic to the HPLC/MS/MS system, and these were kept separate from those in the samples by installation of a second column in between the solvent reservoirs and auto injector. Details of the instrumental set-up and parameters are provided in Supplementary Section. Analytical measures of quality assurance and control; including blank contamination tests, method precision and accuracy, and cross laboratory validation were all satisfactory. Full details are provided in Supplementary Section. 2.3. Pharmacokinetic modeling A simple pharmacokinetic (PK) model was applied to further study the effect of episodic blood loss on serum PFAA concentrations. The PK model is a simple 1-compartment, first order model previously calibrated and used to study accumulation of PFOA and PFOS in the general population in Australia (Thompson et al., 2010). The model has also been used to study exposure of Americans to PFOS and PFOA (Lorber and Egeghy, 2011; Egeghy and Lorber, 2011). At steady state, the simple 1-compartment model can be solved for concentration as:

CðssÞ ¼ ½D=ðkP  VdÞ

ð1Þ

where C(ss) is the steady state concentration of a given PFAA in the serum (ng mL1), D is the constant daily absorbed intake dose (ng kg1 d1), kP is if the first order elimination rate (d1), and Vd is the volume of distribution (mL kg1). Values of 170 mL kg1 and 230 mL kg1 were assigned for the Vd’s of PFOA and PFOS respectively, and elimination rate constants (kP) of 0.0008 d1 for

PFOA and 0.0003 d1 for PFOS were used. These parameters were calibrated using real world data (Thompson et al., 2010). Then, given the Australian male population blood concentrations of 5.9 ng mL1 for PFOA and 16.6 ng mL1 for PFOS, we rearranged Eq. (1) to estimate general male population PFOA and PFOS intakes of 0.80 and 1.15 ng kg1 d1, respectively (Thompson et al., 2010). The model was applied in a non-steady state mode to both PFOA and PFOS in males undergoing venesections, and to PFOA in the study of regular menstrual blood loss in women. These compounds were chosen because they were the most frequently detected compounds, and the necessary model parameters including a background intake estimate, kP, and Vd, were available. The model was applied to male venesection patients, and then to study menstruation as a loss mechanism in females. In constructing the modeling exercise, several assumptions were required. These are discussed in detail in Supplementary Material, including some sensitivity analysis demonstrating the impact of key assumptions. Briefly, they are: (1) there is no aspect of the haemochromatosis blood disorder that influences exposure or toxicokinetics of PFAAs in the body; (2) there are no systematic differences in the demographics of the two groups (venesection patients and general population controls) or other key factors such as sampling time that would cause body burdens to be different in the two populations; (3) PFAA concentrations in the venesection patients were at steady state in 2004, and were the same as the background general population at that time; (4) PFAA intakes were constant during the time of simulation, set at background intakes that have been determined previously and discussed above (Thompson et al., 2010); (5) venesection treatments only remove PFAAs from the blood serum reservoir, not from a full body reservoir of PFAAs; and (6) the female general population intake of PFOA is 0.8 ng/kg d, same as males, and their body burden without an additional removal mechanism would be the same as males. The model was run in an iterative, simulation mode on an ExcelÓ spreadsheet, beginning in 2004 (when venesection data was first available) with initial serum concentrations at background levels. The spreadsheet maintained mass balances of PFOA and PFOS, and concentrations, when needed, were equal to the mass divided by the volume of distribution, Vd. The time interval was 1 month. The only intake was a background intake (determined as days per month * daily intakes) and general elimination from the reservoir was modeled as a first order process (determined as mass at the month’s beginning * elimination rate). Each venesection involved the withdrawal of 450 mL whole blood, or 225 mL of serum. This same volume was withdrawn for each venesection event, regardless of the individual’s gender, weight, age, or other factors. The amount of PFAA mass (ng) removed during each event is calculated as 225 mL * concentration at the end of previous time step of the model (ng mL1). Simulations of PFOA and PFOS were run for each of the 23 composite samples representing male venesections. For each composite, the times and number of venesections, and replenishment time following the last venesection, were input into the model. These details for each simulation are provided in Table 2. Following the last venesection, the blood reservoir was allowed to replenish until the sampling date in 2009. For evaluating menstruation, a simulation started at the body burden found in males at 5.9 ng/mL PFOA in Thompson et al. (2010), with removals of PFOA by menstruation occurring once per month until a new steady state was reached. Assuming a 35 mL blood loss during menstruation, a plausible quantity despite the inherent variability in menses (Fraser et al., 2001), 50% of which is serum, the monthly loss is then modeled as 17.5 mL * serum concentration at the beginning of the previous month (ng/mL).

Table 1 Average concentration [+standard deviation] in ng mL1 of serum PFAAs in male pooled serum samples stratified by age and venesection history. Analyte

Group 1 (G1) >10 venesections since 2004, >365 d between last venesection and sample collection

Group 2 (G2) >10 venesections since 2004, 365 d or less between last venesection and sample collection

Group 3 (G3) 10 or less venesections since 2004, >365 d between last venesection and sample collection

Group 4 (G4) 10 or less venesections since 2004, 365 d or less between last venesection and sample collection

General Public Pooled samples containing 100 donors, collected 2008– 2009 (Toms et al., 2014)a

Males > 60 yrs (61–78 yrs; M1)

PFOA PFNA PFDA PFHxS PFOS

2.9 [0.98] 0.5 [0.01] 0.2 [0.02] 2.3 [0.28] 11.6 [0.21]

n=2

3.1 0.5 0.2 2.2 8.4

n=4

4.2 [1.17] 0.7 [0.18] 0.3 [0.08] 2.1 [0.10] 13.4 [1.41]

n=2

3.6 0.7 0.3 2.2 10.1

n=1

7.3 [0] 1.1 [0] 0.2 [0] 3.1 [0.49] 19.0 [0.28]

n = 2b

Males 60 yrs or less (37–60 yrs; M2)

PFOA PFNA PFDA PFHxS PFOS

3.2 0.7 0.2 1.4 9.8

n=1

2.5 0.5 0.2 4 9.1

n=1

3.2 60 yrs (61–81 yrs)

PFOA PFNA PFDA PFHxS PFOS

n=0

2.5 0.5 0.2 1.6 8.6

n=2

4.7 [0.93] 0.6 [0.05] 0.3 [0.03] 1.9 [0.53] 12.7 [0.64]

n=3

n=0

5.7 [0.42] 1.1 [0] 0.2 [0] 3.3 [0.42] 17.7 [1.27]

n = 2b

3.7 0.5 0.2 2.4 10

n=1

n=1

3.9 1.2 0.3 1.8 9.8

n = 4c

Females 60 and younger (43–59 yrs)

PFOA PFNA PFDA PFHxS PFOS

No sample

[0.62] [0.03] [0.01] [0.86] [0.67]

[0.08] [0.09] [0.02] [1.05] [3.86]

n=0 No sample

n=0 No sample

No sample

3.3 0.8 0.4 3.6 8.6

[0.81] [0.22] [0] [0.66] [2.09]

M. Lorber et al. / Chemosphere 118 (2015) 170–177

Age/gender

a Pools of general population samples subject to different age stratification, with data presented from 2 gender specific pools from each of the following age groups: 31–45 yrs, 46–60 yrs, and >60 yrs. For comparison with the specific venesection patient age groups, the following general public pools were averaged. b >60 yrs. c 31–45 yrs, 46–60 yrs.

173

174

M. Lorber et al. / Chemosphere 118 (2015) 170–177

Table 2 Results of model simulations on each pooled sample, and the input data used. Sample codea

Interval between venesection (d)

Duration of treatment, since 2004 (d)

Number of venesection events

Replenishment time (d)

Measured/Modeled serum PFOA (ng mL1)

Measured/Modeled serum PFOS (ng mL1)

M1G1A M1G1B M1G2A M1G2B M1G2C M1G2D M1G3A M1G3B M1G4A M2G1A M2G2A M2G3A M2G4A M2G4B M2G4C

64 62 53 93 78 75 84 59 129 75 59 80 117 114 41

995 985 1228 1484 1448 1512 557 351 1034 1058 1083 573 875 747 296

16 16 23 16 19 20 7 6 8 14 18 7 8 7 7

610 770 110 131 115 106 819 882 153 793 155 989 81 124 114

3.7/3.8 2.3/4.0 3.0/2.4 2.8/3.0 4.0/2.8 2.6/2.7 3.4/4.7 5.0/4.8 3.6/3.9 3.2/4.8 2.5/2.7 3.2/4.9 3.1/3.6 3.3/3.8 3.7/3.6

11.4/6.9 11.7/7.4 8.4/4.1 9.3/5.9 8.2/5.1 7.6/4.9 12.4/10.7 14.8/11.4 10.1/9.2 9.8/8.1 9.1/5.1 10.2/11.1 10.1/8.8 16.2/9.5 10.2/9.2

3.3/3.7

10.6/7.9

Average a

These sample codes are from Table 1. For example, ‘‘M1G1A’’ includes M1: Males > 60 years, first row; G1: >10 venesections since 2004, >365 d between last venesection and sample collection, first column, A: first of 2 samples in M1G1; second is M1G1B.

3. Results and discussion 3.1. Measured Serum Concentrations Five of the investigated PFAAs were consistently detected in the samples, PFOA, PFNA, PFDA, PFHxS and PFOS (Table 1). The results were consistent with the notion that blood loss will lower body burdens of PFAAs. In general, the male venesection patients had serum concentrations 40–50% less than males of the general public. For PFOS and PFOA, for example, the overall weighted average concentrations of the general population males determined from the pooled results in Table 1 were 17.7 and 6.3 ng mL1 respectively, while they were 10.6 and 3.3 ng mL1 respectively, in the venesection subpopulation. This equates to a difference of 40% and 48% for PFOS and PFOA respectively. The difference between the female venesection patients and the corresponding general population was less pronounced. The overall weighted average of PFOS and PFOA for all females in the general population was 12.4 and 4.5 ng mL1, respectively, while it was 10.6 and 3.7 ng/mL, respectively, in the venesection patients. This equates to a difference of 15% and 18% for PFOS and PFOA, respectively. The only detected PFAA that was not consistently lower in the venesection population was PFDA, which gave average results in male composites equivalent to the concentrations measured in the general public, in the 0.2–0.3 ng mL1 range, which is near the detection limit for PFDA of 0.07 ng mL1. This is potentially due to low exposures resulting in concentrations of this compound that are closer to the limits of quantification, and so further detail is obscured by an inability to measure with greater accuracy. Alternatively, the serum half-life for PFDA may be shorter than that of other PFAAs, making elimination via venesection less relevant. Comparison of males and females in the general population reinforces the typical gender differences described in the introduc-

tion. For the pooled samples brought into this analysis from Toms et al. (2014), the descriptions above note higher PFOS and PFOA concentrations in males compared to females. The female pools had PFOA and PFOA concentrations of 4.5 and 12.4 ng mL1, respectively, which is about 29% and 30% lower than the corresponding male general population pools, 6.3 and 17.7 ng mL1. This comparison considers both age ranges. As described in the introduction, the differences between male and females become smaller or are nonexistent at older ages. For the >60 years pools, male concentrations are higher than females, at 21% higher for PFOA and 7% higher for PFOS. For data exploratory purposes further analysis of venesection group data was performed, investigating differences associated with the number of venesection, replenishment time, and age variables. One might expect these trends relating to venesections: more venesections would lead to lower serum concentrations, and longer replenishment times would lead to higher serum concentrations as the body is given more time to regain steady state. For this analysis, we focused on the 15 male composites only. While generally the number of composites is low, there are more composites in the males, and as seen in Table 1, all groupings are represented by at least one sample for males, while there were no composites for 4 of the 8 groupings for females. The Wilcoxon rank sum test was applied to groupings of the male venesection composites to determine whether these various groupings originated from different populations, from a statistical standpoint. The results of this analysis are shown in Table 3. Grouping data according to number of venesections only (>10 or 610) and ignoring replenishment time and age showed a significant difference for PFOS with a p-value of 0.02 and nearly a significant difference with PFOA at a p-value of 0.07. As would be expected, lower concentrations were found in groups with >10 venesections. The reciprocal analysis, grouping by replenishment time (>365 d 6365 d) and

Table 3 Application of the Wilcoxon rank sum test to the PFOS and PFOA results of the male venesection composite samples to investigate the influence of number of venesections, time of replenishment, and age (average concentrations in ng mL1 provided first, with number of composites, n, in parenthesis). Description of test

PFOA

p-Value

PFOS

Number of venesections

>10 3.0 (8)

610 3.6 (7)

0.07

>10 9.4 (8)

610 11.9 (7)

p-Value 0.02

Days of replenishment

>365 3.5 (6)

6365 3.2 (9)

0.60

>365 11.6 (6)

6365 9.9 (9)

0.04

Age

>60 3.4 (9)

660 3.2 (6)

0.77

>60 10.4 (9)

660 10.8 (6)

0.86

175

M. Lorber et al. / Chemosphere 118 (2015) 170–177

ignoring number of venesections showed a statistically significant difference for PFOS at a p-value of 0.04, but not for PFOA with a pvalue of 0.60. Looking only at age and ignoring number of venesections and replenishment time showed no difference between the two groupings of venesection patients. In the general population data there was an apparent increase in PFOS concentrations with age. As seen in Table 1, lower concentrations are seen in younger males for PFOA (7.3 ng mL1 for males over 60 compared to 5.8 ng mL1 for males under 60) and PFOS (19.0 versus 17.0 ng mL1), but not for the other PFAAs. This general population data is part of a larger data set of PFAA surveys across multiple age groups and over multiple years (Toms et al., 2014). This age trend was not consistent across all survey years (it was absent from the data for 2002/03). Toms et al. (2014) suggest that these inconsistencies in age trends between survey years may be due to the changing degrees of exposure over time. With a reduction in PFAA use in recent years, it seems reasonable to assume that older individuals may have been more exposed to these products. This analysis suggests that a key factor determining the impact to body burden is removal of PFAAs by venesection. The loss of approximately 9% of a person’s circulating serum and the contained PFAAs would be expected to have a significant effect on the concentrations of chemicals remaining in circulation. One might also expect higher concentrations associated with greater replenishment times, since there is more time for the reservoir of PFAAs in the body to recover once the venesections have ceased. This appears to be the case for PFOS, but not for PFOA, even though there was still a small average difference in concentration in the expected direction: PFOA concentrations were 3.5 ng mL1 for longer replenishment times compared to 3.2 ng mL1 for the shorter replenishment time. However the relatively small sample numbers should caution against over interpretation and some of variability in the data may arise from the small number of donors contributing to each composite sample. 3.2. PK modeling results Modeled PFOA serum concentrations over time are also shown for a single composite sample in Fig. 1 showing the effect of venesection and replenishment time on the serum concentrations. The time interval for the model simulation was 1 month, so the results in Fig. 1 associated with a year are actually twelve discrete results connected by a solid line. As seen in Fig. 1, the discrete venesection events, which occurred at most once per a given month, lead to an immediate decline from one monthly concentration to the next. The interval of time between venesections, and more so the replen-

ishment time after the last venesection, shows how the model seeks to return the body burden to the initial steady state. The data used for this figure was from a single composite of male patients, characterized according to our pooling strategy as ‘>60 yrs’, with >10 venesections and >365 d replenishment time; it was specifically the first composite listed in Table 2, with the code M1G1A. The results for each of the 15 simulations of males for PFOA and PFOS, including simulation specifics, are provided in Table 2. The average simulated concentration of PFOA was 3.7 ng mL1, a 37% decline from the initial assumed general population concentration of 5.9 ng mL1. The observed average concentration in the venesection patients was 3.3 ng mL1. Hence for PFOA the model is in very good agreement with the mean of the measured data. For PFOS, the model predicts a concentration drop from 16.6 ng mL1 to 7.8 ng mL1 in the venesection group, a 53% decline, where the mean measured concentration was substantially higher at 10.6 ng mL1. Although the model agreement was poorer for PFOS, overall, the modeling results support the hypothesis that the lower concentrations seen in venesection patients can be explained through the loss of PFAAs via loss of blood during venesection. There appeared to be a reasonable correlation between measured and modeled serum concentrations, although there was up to a 1.5 ng mL1 difference in several individuals for PFOA and up to a 7.3 ng mL1 difference in one individual for PFOS. Overall, the correlation coefficient for PFOA, for observed versus predicted, was 0.43 (R2 = 0.18), and for PFOS, it was 0.68, (R2 = 0.46). The reason why the model seemed to better duplicate the measurements for PFOA on average in comparison to PFOS is unknown. The critical Vd parameter was first calibrated to a value of 170 mL kg1 for PFOA based on human data from a contamination site, and then extended to 230 mL kg1 for PFOS based on animal data (see Thompson et al., 2010, for more detail). At least in terms of model performance, whether a better characterization of reality or not, lower values of Vd in the PFOS simulations would result in higher predictions of final blood concentrations. A sensitivity test was run on the PFOS simulations by changing the Vd from 230 to 170 mL kg1, and it was found that blood concentration predictions for the venesections patients increased by about 30%. This is more in line with observations. The simulation of the effect of menstruation on serum concentrations of PFOA is shown in Fig. 2. The body burden begins at 5.9 ng mL1, the presumption of what it might be without the menstrual loss. It is seen that steady state is regained after about 8 years at a new lower body burden of 4.6 ng mL1, a 22% reduction. This is in remarkably good agreement with the observed mean body burden in women in the background population of

7.00

PFOA serum concentraon, ng mL-1

PFOA serum concentraon, ng mL-1

7.00 6.00

Number of venesecons = 16 Recover me = 610 days

5.00 4.00 3.00 2.00 1.00 1

2

3

4

Years from start of venesecon

6.00 5.00 4.00 3.00 2.00 1.00 0.00

0

1

2

3

4

5

6

7

8

9

10

Years from start of menstruaon Fig. 1. Modeled serum PFOA concentration profile with time for a single composite sample.

Fig. 2. Modeled reductions in serum PFOA concentration due to menstruation.

176

M. Lorber et al. / Chemosphere 118 (2015) 170–177

4.5 ng mL1. This value is the average of four pooled samples corresponding to 2008/9 reported in Toms et al. (2014). The model does show some response to changes in the assumed loss of blood by menstruation. This solution assumed a monthly loss of 35 mL of blood, or 17.5 mL of serum assuming serum is 50% of blood by weight. If instead the assumed monthly serum loss was 35 mL (twice that assumed) or 9 mL (about half assumed), the resulting simulated body burden at steady state would be 3.8 and 5.2 ng mL1, respectively. Though not a definitive proof, this modeling exercise provides supportive evidence for the finding that menstrual losses could explain the difference in male and female body burdens. This finding is supported by Knox et al. (2011), who studied the relationship between higher levels of PFAAs in a cohort of greater than 69 000 women who had been exposed to PFAAs by proximity to a contaminated water supply in West Virginia, USA, and early onset of menopause. While they found evidence for early onset of menopause in these exposed women, they also found that women who had hysterectomies had higher levels of PFAAs as compared to women who did not have hysterectomies. This suggests that menstruation served as a mechanism for removal of PFAAs in these individuals, and supports the hypothesis in this study. Using population-based pharmacokinetic modeling, Wong et al. (2014) demonstrated enhanced elimination of PFOS in women (of menstrual age) as compared to men, using data from the US NHANES from 1999 to 2012, and they attributed this enhanced elimination to menstruation. 4. Conclusions We sought to understand the role blood loss plays in lowering serum PFAA concentrations in humans. Trends in two sets of populations were examined. One was a cohort of individuals who had undergone blood removal treatments for medical reasons, and the other is females of the general population, who experience a regular loss of blood through menstruation. Measurements from both sets of individuals show lower concentrations of PFAAs that we hypothesize are associated with blood removals or loss. We found that the modeled concentrations mirrored these measured trends relatively well. Uncertainties identified include the relatively small number of samples from the venesection population, the simplicity of the PK model, and the relatively small amount of data that went into calibration of the model parameters. Uncertain model parameters include the overall rates of elimination, which were determined from a single study on an occupational exposure cohort, and the volumes of distribution, which were calibrated in a modeling exercise applied to a contamination site (Thompson et al., 2010). Given that the only expectation we have for this model is that it might provide a first look at how a serum concentration can change as a result of a change in a possible elimination mechanism, we feel it is an appropriate choice of model and an appropriate application. Like all such modeling exercises, more data to develop and test the model is desired. We note that the differences in male and female concentrations of the PFAAs may have more to do with differences in exposure patterns and/or differences in disposition of PFAAs following exposure. Overall, this study provides data and modeling that supports the initial hypothesis that ongoing blood loss explains lower PFAA concentrations in humans. More data and modeling evidence would, of course, better support this hypothesis. Disclaimer The views expressed in this article are those of the authors and do not necessarily reflect the views or policies of the U.S. Environ-

mental Protection Agency. Mention of trade names or commercial products does not constitute endorsement or recommendation for use. Acknowledgements J.F. Mueller is funded by Australian Research Council (ARC) Future Fellowship (FF120100546). L.-M.L. Toms is funded by an ARC Discovery Early Career Researcher Award (DE120100161). The National Research Centre for Environmental Toxicology is co-funded by Queensland Health. The authors are grateful for the statistical support provided by John Fox of the United States Environmental Protection Agency. Appendix A. Supplementary material Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.chemosphere. 2014.07.093. References Calafat, A.M., Kuklenyik, Z., Reidy, J.A., Caudill, S.P., Tully, J.S., Needham, L.L., 2007. Serum concentrations of 11 perfluoroalkyl compounds in the U.S. population: data from the national health and nutrition examination survey (NHANES) 2003–2004 and comparisons with NHANES 1999–2000. Environ. Sci. Technol. 41 (7), 2237–2242. Egeghy, P.P., Lorber, M., 2011. An assessment of the exposure of Americans to perfluorooctane sulfonate: a comparison of estimated intake with values inferred from NHANES data. J. Exp. Sci. Environ. Epi. 21 (2), 150–168. Fraser, I.S., Warner, P., Marantos, P.A., 2001. Estimating menstrual blood loss in women with normal and excessive menstrual fluid volume. Obstet. Gynecol. 98, 806–814 (5, Part 1). Fromme, H., Tittlemier, S.A., Volkel, W., Wilhelm, M., Twardella, D., 2009. Perfluorinated compounds – exposure assessment for the general population in western countries. Int. J. Hydrogen Environ. Health 212 (3), 239–270. Harada, K.I.K., Morikawa, A., Yoshinaga, T., Saito, N., Koizumi, A., 2005. Renal clearance of perfluorooctane sulfonate and perfluorooctanoate in humans and their species-specific excretion. Environ. Res. 99, 253–261. Houde, M., Martin, J.W., Letcher, R.J., Solomon, K.R., Muir, D.C.G., 2006. Biological monitoring of polyfluoroalkyl substances: a review. Environ. Sci. Technol. 40 (11), 3463–3473. Ingelido, A.M., Marra, V., Abballe, A., Valentini, S., Iacovella, N., Barbieri, P., et al., 2010. Perfluorooctanesulfonate and perfluorooctanoic acid exposures of the Italian general population. Chemosphere 80, 1125–1130. Knox, S.S., Jackson, T., Javins, B., Frisbee, S.J., Shankar, A., Ducatman, A.M., 2011. Implications of early menopause in women exposed to perfluorocarbons. J .Clin. Endocrinol. Metab. 96 (6), 1747–1753. Kubwabo, C., Vais, N., Benoit, F., 2004. A pilot study on the determination of perfluorooctanesulfonate and other perfluorinated compounds in blood of Canadians. J. Environ. Monit. 6, 540–545. Lorber, M., Egeghy, P.P., 2011. Simple intake and pharmacokinetic modeling to characterize exposure of Americans to perfluorooctanoic acid PFOA. Environ. Sci. Technol. 45 (19), 8006–8014. NICNAS, 2007. NICNAS alert no. 6 Perfluorooctanoic acid (PFOA) & Perfluorocarboxylic acid (PFCA). DoHa. Ageing. Sydney, Australian Government: 2. . NICNAS, 2008. NICNAS alert no. 8 Perfluorooctane sulfonate (PFOS) & perfluoroalkyl sulfonate (PFAS). 2008. DoHA. Sydney, NICNAS: 1–5. . OECD, 2002. Co-operation on existing chemicals: Hazard assessment of perfluorooctane sulfonate (PFOS) and its salts. 21-Nov-2002. Organization for Co-operation and Development. ENV/JM/RD (2002) 17/FINAL. Olsen, G.W., Huang, H.-Y., Helzlsouer, K.J., Hansen, K.J., Butenhoff, J.L., Mandel, J.H., 2005. Historical comparison of perfluorooctanesulfonate, perfluorooctanoate, and other fluorochemicals in human blood. Environ. Health Perspect. 113 (5), 539–545. Prevedouros, K.C., Buck, R.C., Korzieniowski, S.H., 2006. Sources, fate and transport of perfluorocarboxylates. Environ. Sci. Technol. 40 (1), 32–44. Silvaggio, T., Mattison, D.R., 1994. Setting occupational health standards: toxicokinetic differences between men and women. J. Occup. Med. 36 (8), 849–854. Soldin, O.P., Mattison, D.R., 2009. Sex differences in pharmacokinetics and pharmacodynamics. Clin. Pharmacokinet. 48 (3), 143–157. Stockholm Convention, 2010. Stockholm Convention on Persistent Organic Pollutants (POPs). Stockholm Convention, 2010. (retrieved 20.08.10). Thompson, J., Lorber, M., Toms, L.-M.L., Kato, K., Calafat, A.M., Mueller, J.F., 2010. Use of simple pharmacokinetic modeling to characterize exposure of Australians to perfluorooctanoic acid and perfluorooctane sulfonic acid. Environ. Int. 36 (4), 390–397. Toms, L.-M.L., Calafat, A., Kato, K., Thompson, J., Harden, F., Hobson, P., Sjödin, A., Mueller, J.F., 2009. Polyfluoroalkyl chemicals (PFCs) in human blood serum from children and adults in Australia. Environ. Sci. Technol. 43 (11), 4194–4199. Toms, L.-M.L., Thompson, J., Rotander, A., Hobson, P., Calafat, A.M., Kato, K., Ye, X., Broomhall, S., Harden, F., Mueller, J.F., 2014. Decline in perfluorooctane sulfonate and perfluorooctanoate serum concentrations in an Australian population from 2002 to 2011. Environ. Int. 71, 74–80.

177

Wang, T., Wang, Y., Liao, C., Cai, Y., Jiang, G., 2009. Perspectives on the Inclusion of perfluorooctane sulfonate into the Stockholm Convention on persistent organic pollutants. Environ. Sci. Technol. 43 (14), 5171–5175. Wong, F., MacLeod, M., Mueller, J.F., Cousins, I.T., 2014. Enhanced elimination of perfluorooctane sulfonic acid by menstruating women: evidence from population-based pharmacokinetic modeling. Environ. Sci. Technol. (Just accepted manuscript 2014, http://dx.doi.org/10.1021/es500796y Publication Date (Web): June 19, 2014). Yeung, L.W.Y., Taniyasu, S., Kannan, K., Xu, D.Z.Y., Guruge, K.S., Lam, P.K.S., Yamashita, N., 2009. An analytical method for the determination of perfluorinated compounds in whole blood using acetonitrile and solid phase extraction methods. J. Chromatogr., A 1216, 4950–4956.