Science of the Total Environment 625 (2018) 566–574

Contents lists available at ScienceDirect

Science of the Total Environment

journal homepage: www.elsevier.com/locate/scitotenv

PFOA is associated with diabetes and metabolic alteration in US men:

National Health and Nutrition Examination Survey 2003–2012
Xiaowei He a,1, Yuanxin Liu a,1, Bo Xu b, Liubao Gu a, Wei Tang a,⁎
a
Department of Endocrinology, Islet Cell Senescense and Function Research Laboratory, Jiangsu Province Geriatric Institute, 30 Luojia Road, Nanjing, Jiangsu 210024, China
b
School of Public Health, Nanjing Medical University, Nanjing 211166, China

H I G H L I G H T S G R A P H I C A L A B S T R A C T

• 7904 adults were used to examine asso-

ciations of PFAS with diabetes and me-
tabolite levels.
• Serum PFOA was positively associated
with diabetes mellitus in men.
• PFOA may disrupt cholesterol metabo-
lism at environmentally exposures.

a r t i c l e i n f o a b s t r a c t

Article history: Exposure to perfluoroalkyl substances (PFAS) is associated with a range of adverse health effects. However, it re-
Received 24 September 2017 mains unclear whether PFAS at environmentally relevant exposure levels are related to diabetes and metabolite con-
Received in revised form 16 December 2017 centrations in adults. Using cross-sectional data from 7904 adults (age ≥ 20 years) in the 2003–2012 National Health
Accepted 17 December 2017
and Nutrition Examination Survey (NHANES), we examined the association of PFAS with the prevalence of diabetes
Available online xxxx
and metabolite concentrations. A multivariate logistic regression was applied to investigate the associations of dia-
Editor: D. Barcelo betes prevalence with serum perfluorooctanoate (PFOA), perfluorooctane sulfonate (PFOS), perfluorohexane sulfo-
nate (PFHxS) and perfluorononanoate (PFNA) levels. A multivariate generalised linear regression was further
Keywords: performed to investigate the associations between PFAS exposure and some metabolites. We identified a strong pos-
Perfluoroalkyl substances itive association between serum PFOA and diabetes prevalence in men with an adjusted model (OR: 2.66, 95% CI:
Perfluorooctanoate 1.63–4.35; P for trend = 0.001). No significant association between serum PFOA and diabetes prevalence was ob-
Diabetes served in women (OR: 1.47, 95% CI: 0.88–2.46; P for trend = 0.737). Furthermore, diabetes was not related to
Metabolites PFOS, PFHxS and PFNA, regardless of gender. In the gender-stratified generalised linear models, men and women
Total cholesterol
with the highest PFOA levels demonstrated a 1.43% (95% CI: 0.62%–2.34%) and a 1.07% (95% CI: 0.27%–1.97%) greater
increase in serum total cholesterol (P for trend = 0.006 and 0.001) compared to those with the lowest PFOA levels.
There were no significant associations between serum PFOA and other metabolites. These results provide epidemi-
ological evidence that environment-related levels of serum PFOA may be positively associated with the prevalence
of diabetes in men and with total cholesterol in adults. Further clinical and animal studies are urgently needed to
elucidate putative causal relationships and shed light on the potential mode of action involved.
© 2017 Published by Elsevier B.V.

⁎ Corresponding author.
E-mail address: drtangwei@aliyun.com (W. Tang).
1
These authors contributed equally to this work.

https://doi.org/10.1016/j.scitotenv.2017.12.186
0048-9697/© 2017 Published by Elsevier B.V.
 X. He et al. / Science of the Total Environment 625 (2018) 566–574
1. Introduction
Diabetes mellitus, usually referred to as diabetes, is a group of meta-
bolic disorders characterised by high blood glucose levels over a
prolonged period (Moyer et al., 2012). Without effective management,
diabetes can induce many health complications. Recent estimates attri-
bute almost 5.1 million deaths to diabetes in adults, representing 8.4% of
all-cause deaths globally (IDF Diabetes Atlas Group, 2015). Risk factors
for diabetes include genetic predisposition, ethnic background, age,
lack of physical exercise, obesity and use of tobacco (Karnes et al.,
2014; Thiering et al., 2011). Recent research has indicated that environ-
mental exposure could have contributed to the recent dramatic surge in
diabetes (Bond and Dietrich, 2017; Honda et al., 2017). Environmental
chemicals used in most of today's family households may have
endocrine-disrupting properties and thus disrupt the endogenous hor-
mone system (Audouze et al., 2013) (Gore et al., 2015). These chemicals
may cause impaired insulin secretion and insulin sensitivity, causing the
development of diabetes (Domazet et al., 2016).
Perfluoroalkyl substances (PFAS) are human-made compounds that
have been used in a multitude of industrial and commercial applica-
tions, including fire extinguishing foams, photographic emulsifiers and
stain-, grease- and water-resistant protection for clothing, furniture,
carpet, leather, and paper (Calafat et al., 2007; Lau et al., 2007). PFAS
contain a group of substances that are classified as persistent organic
pollutants due to their strong resistance to biodegeneration, hydrolysis,
photolysis and atmospheric photooxidation (Domazet et al., 2016). The
production of perfluorooctane sulfonate (PFOS) and some other PFAS
has been voluntarily discontinued in the United States (Buck et al.,
2011). However, due to its ubiquity and persistence, PFAS still
Fig. 1. Flow chart depicting the final analysed samples for each outcome from five successive waves of NHANES cycle (2003–2012). NHANES, National Health and Nutrition Examination
Survey; PFAS, perfluoroalkyl substances; FPG, fasting plasma glucose; Fins, Fasting insulin; HOMA-IR, homeostasis model assessment of insulin resistance; TC, total cholesterol; HDL-C,
HDL-cholesterol; LDL-C, LDL-cholesterol; TG, triglycerides.
567
contaminate various environmental media (Butt et al., 2010). Mean-
while, biomonitoring of the general population has shown widespread
exposure to certain PFAS (Christensen et al., 2016). Of note, mean
serum half-lives in humans are estimated to be 4.8 years for PFOS,
3.5 years for perfluorooctanoate (PFOA) (Olsen et al., 2007) and
8.5 years for perfluorohexane sulfonate (PFHxS). Once absorbed, these
chemicals can persist in the body by binding to liver and serum proteins
(Ng and Hungerbuehler, 2015). Furthermore, the widely utilised newer
"replacement" of PFAS ensures a sustained exposure risk (Wang et al.,
2013).
High serum PFOA level was found among workers with occupational
exposure to PFOA at a DuPont chemical plant (Steenland and Woskie,
2012). Workers exposed to PFOA showed a twofold increase in diabetes
mortality in comparison to other non-exposed regional workers
(Leonard et al., 2008; Steenland and Woskie, 2012). However, studies
of residents in this community indicate a null association between
PFOA exposure and type II diabetes, both cross-sectionally and prospec-
tively (Karnes et al., 2014; MacNeil et al., 2009). Notably, exposure
levels in these workers or residents are much higher than in general
US populations of non-exposed region.
As a review by Post et al. mentions (Post et al., 2012), this study's
dose-response curve presented steeper slopes at low levels of PFOA ex-
posure than at high levels of PFOA exposure. This suggests the great im-
portance of exploring low-dose effects in the general population.
Nevertheless, the role of PFOA in prevalent diabetes is still largely un-
known; and inconsistent relationships were observed in the studies
on the general population.
To further clarify the relationship between PFAS and diabetes prev-
alence, we performed a cross-sectional analysis of the 2003–2012
568 X. He et al. / Science of the Total Environment 625 (2018) 566–574

National Health and Nutrition Examination Survey (NHANES) to exam- were included in this analysis. Concentrations below the limit of detec-
ine the association of serum PFAS levels with diagnosed diabetes in tion (LOD) were reported by NHANES as the LOD divided by the square
adults, stratified by gender. root of two.

2. Research design and methods 2.5. Potential confounders

2.1. Study population
Publicly available data for participants who enrolled in this study
were merged from five cycles of the NHANES (2003–2004,
2005–2006, 2007–2008, 2009–2010, and 2011–2012). Detailed infor-
mation of the survey design and methods are available on the
NHANES website [Centers for Disease Control and Prevention (CDC)].
Briefly, the NHANES is a series of continuous, multicomponent cross-
sectional surveys designed to be nationally representative of the non-
institutionalised U.S. civilian population. After providing written in-
formed consent, each participant visited a mobile examination center
(MEC) for physical assessment, examination, and biological sample
collection.
Individuals participating in the NHANES over a range of 10 years
were selected to from a random representative subset for the measure-
ment of serum PFAS and serum metabolite levels. Of 24,146 adults
≥20 years of age who engaged in both the NHANES interview and the
examination, those who had missing PFAS information or were current-
ly pregnant were excluded. In a final sample of 7904 participants, we in-
vestigated the association of serum PFAS levels with diabetes
prevalence and metabolite concentrations. Due to the self-reported na-
ture of this assessment in NHANES, we cannot identify the type of diabe-
The following variables were chosen a priori as covariates in the
models because of their biological relevance to the outcomes: age,
race, body mass index (BMI), education level, energy intake, serum co-
tinine (a exposure marker of environmental cigarette smoke), alcohol
consumption, poverty-income ratio (PIR; family income divided by
the federal poverty threshold controlling for inflation and family size),
and time spent in front of a screen (watching television, playing video
games and using computers; measured as hours per day in the forego-
ing 30 days) (Goldfield et al., 2013; Holst et al., 2017; Jarvandi et al.,
2015). The information of these variables was extracted from the
NHANES questionnaires. In addition, a variable was considered as a con-
founder if its inclusion in the model could change the beta coefficient for
any quartiles of the PFAS exposure by N 10% (James-Todd et al., 2012).
2.6. Statistical analysis
Serum concentrations of PFAS were categorized in quartiles on the
basis of the sample distribution. We applied a multivariate logistic
Table 1
Characteristics of 2003–2012 NHANES study participants and distribution of serum
perfluoroalkyl substances (PFAS) by gender.a

tes. However, it must be pointed out that the vast majority of Characteristics Male Female
participants with diabetes in NHANES are likely to be type II in origin, (n = 3956) (n = 3948)
since this is a nationally general population study among adults in US Age (years) 50.22 ± 0.29 49.33 ± 0.29
(Bainbridge et al., 2008; Menke et al., 2013). Final analysed samples BMI (kg/m2) 28.45 ± 0.10 29.22 ± 0.12
for each outcome among eligible populations are summarised in Fig. 1. Energy intake (kcal) 2375.66 ± 15.23 1744.19 ± 10.40
Serum cotinine (ng/ml) 71.77 ± 2.25 42.41 ± 1.69
Race (%)
2.2. Primary outcome: diabetes mellitus Mexican American 16.8 17.0
Other Hispanic 7.2 8.6
Participants were asked by NHANES interviewers if they had ever Non-Hispanic White 48.4 47.2
been told by a doctor or other health professional that they had diabetes Non-Hispanic Black 20.6 20.4
Other race - including multi-racial 7.1 6.8
or sugar diabetes (other than during pregnancy). Based on answering Education level (%)
“yes” to this question, participants were coded as having been diag- Less than 9th grade 13.9 11.8
nosed with diabetes. 9–11th grade 15.3 15.5
(includes 12th grade with no diploma)
High school grad/GED or equivalent 23.3 22.5
2.3. Secondary outcome: metabolite concentrations
Some college or AA degree 26.0 29.6
College graduate or above 21.2 20.4
To observe whether there are relationships between PFAS and me- Time spent in front of a screen (%)
tabolism alteration, we conducted additional analyses in a random sub- ≤2 h 41.9 45.3
group with metabolite data. We explored the associations between N2 h 58.1 54.7
PIR (%)
PFAS with markers of glucose metabolism [fasting plasma glucose ≤1 17.1 20.2
(FPG), fasting insulin levels (Fins), the homeostasis model assessment N1 75.0 72.0
of insulin resistance (HOMA-IR)] and lipid metabolism [total cholesterol Alcohol usage (%)
(TC), triglycerides (TG), high-density lipoprotein cholesterol (HDL-C), Yes 76.9 53.4
No 15.6 36.7
low-density lipoprotein cholesterol (LDL-C)] and blood pressure. Ab-
BMI (%)
normalities of these metabolic parameters are known to increase the b25 28.4 31.5
risk for diabetes (Lorenzo et al., 2007; Z. Zhang et al., 2014). 25–30 38.8 28.7
Detailed information on sample collection and treatment at the U.S. N30 31.4 38.3
CDC are available in the previous studies (Peng et al., 2015; Xu et al., PFOA (ng/ml) 4.50 ± 0.06 3.46 ± 0.04
PFOS (ng/ml) 20.80 ± 0.32 14.51 ± 0.26
2015).
PFHxS (ng/ml) 2.88 ± 0.05 1.94 ± 0.04
PFNA (ng/ml) 1.52 ± 0.02 1.30 ± 0.03
2.4. Primary exposure: serum PFAS levels
Age, BMI, energy intake, serum cotinine and PFAS are presented as mean values and their
standard errors.
Briefly, serum PFAS levels were analysed at the U.S. CDC using online Race, education level, time spent in front of a screen, alcohol usage and BMI are presented
solid-phase extraction–high performance liquid chromatography– as percentages.
turbo ion spray-tandem mass spectrometry (SPE-HPLC-TIS-MS/MS), BMI, body mass index; PIR, poverty-Income ratio; PFOA, perfluorooctanoic acid;
PFOS, perfluorooctane sulfonic acid; PFHxS, perfluorohexane sulfonic acid; PFNA,
as reported previously (Calafat et al., 2007; Jain, 2014). perfluorononanoic acid.
Serum concentrations of four PFAS (PFOA, PFOS, PFHxS, and PFNA), a
Subjects were excluded if they had a positive laboratory pregnancy test or self-re-
which were detectable in N 98% of the 2003–2012 survey participants, ported to be pregnant at exam (n = 202).
 X. He et al. / Science of the Total Environment 625 (2018) 566–574 569

Table 2 (IBM Corp., Armonk, NY, USA) was applied for all statistical analyses.
Distribution of diabetes outcome and concentrations of serum metabolites in the A P value b 0.05 was designated as the criteria for statistical significance.
participants.

Characteristics Male Female 3. Results

n Mean ± SE (%) n Mean ± SE (%)
3.1. Baseline characteristics
Diabetes (%) 3874 11.8 3865 11.6
FPG (mg/dL) 1918 111.30 ± 0.87 1928 104.81 ± 0.77
Fins (uU/mL) 1909 13.98 ± 0.47 1917 12.50 ± 0.27 Table 1 shows the general characteristics and mean concentrations
HOMA-IR 1907 4.12 ± 0.17 1914 3.41 ± 0.09 of serum PFAS levels among the participants included in the current
TC (mg/dL) 3954 193.5 ± 0.67 3947 200.47 ± 0.66 study from the NHANES 2003–2012 database. Table 2 presents the dis-
HDL-C (mg/dL) 3954 48.32 ± 0.23 3946 57.86 ± 0.26
LDL-C (mg/dL) 1844 115.28 ± 0.84 1903 116.70 ± 0.83
tribution of the diabetes outcome and serum metabolite levels of the
TG (mg/dL) 1916 148.58 ± 3.00 1927 124.47 ± 2.34 participants. Diabetes mellitus was reported in 11.8% of men and
SBP (mm Hg) 3804 125.15 ± 0.28 3739 123.44 ± 0.35 11.6% of women in this study population.
DBP (mm Hg) 3804 71.18 ± 0.21 3739 68.73 ± 0.22

FPG, fasting plasma glucose; Fins, fasting insulin; HOMA-IR, homeostasis model assessment
of insulin resistance; TC, total cholesterol; HDL-C, HDL-cholesterol; LDL-C, LDL-cholester-
ol; TG, triglyceride; SBP, systolic blood pressure; DBP, diastolic blood pressure.
regression analysis to investigate the association of a diabetes outcome
with each PFAS separately. As gender is a particularly crucial
predicator, we conducted a separate analysis stratified by gender.
To evaluate the additive effect of four PFAS, we categorized each
PFAS into 10 groups using 10th percentiles. The cumulative PFAS
level was made by adding up the category number of each PFAS,
resulting in a value of 4–40. Then the cumulative PFAS was catego-
rized into quartiles, making four groups (Moon, 2014). We showed
effect estimates and their matching 95% confidence intervals (CIs)
for each quartile compared with the lowest quartile (the reference
group). P for trend was performed by considering the PFAS category
as a linear variable in the models.
A generalised linear model was performed to estimate the associa-
tions between increases in the interquartile ratio (IQ ratio = 75th/
25th percentiles of PFAS level) for levels of metabolites and PFAS. Con-
tinuous outcomes were log-transformed to approximate a normal dis-
tribution. Gender was used as the stratified variable to conduct a
separate analysis. Testing for linear trends was performed by entering
the PFAS quartiles as a continuous variable into the same model. The
magnitudes of association were presented as the average percentage
difference (% diff) in metabolic levels for comparison of interquartile ra-
tios (IQ ratio) for PFOA, calculated as [(IQ ratiobeta) − 1] × 100. This
method scales the magnitudes of association to an exposure range,
which existed in both genders of the study population.
All regression models were conducted after adjusting for age, race,
BMI, education levels, energy intake, serum cotinine, alcohol consump-
tion, PIR, and the time in front of a screen (watching television, playing
video games or using computers). The IBM SPSS software, version 20.0
3.2. Diabetes mellitus
The primary effect estimates from the multinomial logistic regres-
sion models adjusted for a group of covariates for individual and cumu-
lative PFAS are presented in Table 3. Notably, significant positive
associations were observed between diabetes prevalence and quartile
2 (Q2; OR: 2.13, 95% CI: 1.30, 3.46), Q3 (OR: 2.44, 95% CI: 1.49, 3.98)
and Q4 (OR: 2.67, 95% CI: 1.63, 4.38) of PFOA, relative to the lowest
Q1 of exposure (P for trend = 0.001) in men. Additionally, there was
a slight association between diabetes prevalence and each quartile of
PFOA in women, although the dose-response relationship in women
was less strong and clear than in men (Fig. 2A). Furthermore, diabetes
prevalence was not significantly associated with any of the categorical
PFOS, PFHxS, PFNA, or cumulative PFAS exposure, regardless of gender.
3.3. Metabolism alteration
Because we found a significant association between PFOA exposure
and the prevalence of diabetes, we further explored the association be-
tween some serum metabolites and PFOA exposure. Estimated % diff
and 95% CI in serum TC concentrations for each IQ ratio increase in
serum PFOA levels are presented separately for men and women in
Table 4 and Fig. 2B. Men with the highest PFOA levels demonstrated a
1.43% (95% CI: 0.62%, 2.34%) greater increase in serum TC (P for trend
b 0.001) than participants with the lowest serum PFOA levels. Women
in Q3 and Q4 of PFOA levels showed a 1.16% (95% CI: 0.44%, 1.97%)
and 1.07% (95% CI: 0.27%, 1.97%) greater increase in serum TC (P for
trend = 0.001) than those participants in Q1 of PFOA levels. We found
no association between PFOA concentration and FPG, Fins, HOMA-IR,
HDL-C, LDL-C, SBP or DBP (Supplementary materials). The relationships
between the biomarkers and PFOS, PFHxS or PFNA are presented in the
Supplementary materials.

Table 3
Odd ratio (OR) for the associations between the quartiles of serum PFAS and diabetes prevalence.

PFAS Gender Q1 Q2 Q3 Q4 P for trend

PFOA Male Reference 2.13a (1.30, 3.46) 2.44a (1.49, 3.98) 2.67a (1.63, 4.38) 0.001
PFOS Reference 1.32 (0.71, 2.45) 1.64 (0.92, 2.93) 1.75 (1.00, 3.04) 0.305
PFHxS Reference 1.99 (1.19, 3.33) 1.87 (1.15, 3.05) 2.31 (1.37, 3.91) 0.071
PFNA Reference 1.25 (0.77, 2.04) 1.17 (0.74, 1.87) 1.19 (0.73, 1.95) 0.457
Cumulative PFAS Reference 1.18 (0.84, 1.65) 1.34 (0.97, 1.86) 1.55 (1.11, 2.14) 0.363
PFOA Female Reference 1.38 (0.87, 2.19) 1.57 (0.95, 2.61) 1.47 (0.87, 2.48) 0.737
PFOS Reference 1.11 (0.66, 1.88) 1.06 (0.63, 1.78) 1.41 (0.82, 2.41) 0.829
PFHxS Reference 0.65 (0.41, 1.03) 0.87 (0.52, 1.43) 1.22 (0.71, 2.11) 0.056
PFNA Reference 0.98 (0.63, 1.54) 1.50 (0.88, 2.57) 1.01 (0.62, 1.65) 0.250
Cumulative PFAS Reference 1.02 (0.75, 1.38) 1.17 (0.84, 1.63) 1.13 (0.80, 1.59) 0.162

Multinomial logistic regression models were used, adjusted for age, race, BMI, education level, energy intake, serum cotinine, alcohol consumption, PIR and the time in front of TV, video
game and computer.
Quartiles (Q) of PFOA (ng/ml): Q1: b2.1; Q2: 2.1–3.34; Q3: 3.34–5.1; Q4: N5.1.
Quartiles (Q) of PFOS (ng/ml): Q1: b7.305; Q2: 7.305–13.1; Q3: 13.1–22.5; Q4: N22.5.
Quartiles (Q) of PFHxS (ng/ml): Q1: b0.9; Q2: 0.9–1.64; Q3: 1.64–2.9; Q4: N2.9.
Quartiles (Q) of PFNA (ng/ml): Q1: b0.738; Q2: 0.738–1.07; Q3: 1.07–1.64; Q4: N1.64.
PFAS, perfluoroalkyl substances; PFOA, perfluorooctanoic acid; PFOS, perfluorooctane sulfonic acid; PFHxS, perfluorohexane sulfonic acid; PFNA, perfluorononanoic acid.
a
P b 0.05.
570 X. He et al. / Science of the Total Environment 625 (2018) 566–574
Fig. 2. Adjusted odd ratio (OR) and 95% confidence intervals (95% CI) for diabetes risk in U.S. adults 2003–2012 by increasing quartiles of serum PFOA levels by gender (A). Estimated
percent difference (% diff) and 95% CI in serum total cholesterol (TC) concentrations in U.S. adults 2003–2012 for each interquartile ratio (IQ ratio) increase in serum PFOA levels by
gender (B). Quartiles (Q) of PFOA (ng/ml): Q1: b2.1; Q2: 2.1–3.34; Q3: 3.34–5.1; Q4: N5.1.
To compare with a previous study demonstrating the association be-
tween PFOA levels and TC in a middle-aged population (Eriksen et al.,
2013), we conducted an effect estimation to examine the relationship
between PFOA and TC, stratified by age and gender. As presented in
Table 5, strong associations were observed in men aged 40–50 (% diff:
2.88%, 95% CI: 0.89%, 4.91%) and 50–60 (% diff: 2.52%, 95% CI: 0.44%,
4.73%), as well as women aged 60–70 (% diff: 2.88%, 95% CI: 0.80%,
5.00%). Scatter plots and fitted lines with 95% CIs of the associations be-
tween PFOA concentrations and serum TC are shown in Fig. 3 for men
aged 40–60 (Fig. 3A) and women aged 60–70 (Fig. 3B).
4. Discussion
Using data from the representative NHANES samples allowed us to
explore the associations between four PFAS and the prevalence of dia-
betes, as well as metabolite concentrations. The most prominent find-
ings were for TC in the overall population and diabetes in men.
A cross-sectional study in an elder Swedish population demonstrat-
ed no significant association between PFOA [median: 3.3, interquartile
range (IQR): 2.5, 4.4 ng/ml] and diabetes prevalence (Lind et al.,
2014). However, in a recent study conducted in 571 working-age
Taiwanese participants, PFOA (median: 8.0, IQR: 5.8, 10.5 ng/ml) was
shown to be inversely associated with the risk of diabetes (Su et al.,
2016). In our study, an adverse impact effect of PFOA (median: 3.3,
IQR: 2.1, 5.1 ng/ml) against diabetes risk was demonstrated. The dis-
crepancy between our discovery and the results from previous studies
may be due to divergent criteria used to define diabetes, different expo-
sure levels of PFAS, different sampling or analysing procedures. In an-
other study conducted with 258 pregnant women, PFOA (median:
3.3 ng/ml) was significantly associated with an increased GDM risk
(Zhang et al., 2015). In the present study, we discovered a significant
gender difference for the associations between PFOA exposure and
Table 4
Estimated percent difference (% diff) and 95% confidence intervals (95% CI) in serum total cholesterol (TC) concentrations in US adults 2003–2012 for each interquartile ratio (IQ ratio)
increase in serum PFOA levels.
TC Males
% diff 95% CI
Q1 Reference
Q2 0.44% −0.44%, 1.34%
Q3 0.80% −0.09%, 1.70%
Q4 1.43%a 0.62%, 2.34%
TC was Log-transformed in models.
% diff = [(IQ ratiobeta) − 1] ∗ 100.
IQ ratio = 75th/25th percentiles of serum PFOA: 2.43.
Quartiles (Q) of PFOA (ng/ml): Q1: b2.1; Q2: 2.1–3.34; Q3: 3.34–5.1; Q4: N5.1.
PFOA, perfluorooctanoic acid.
a
P b 0.05.
diabetes prevalence in adults. To our knowledge, this is the first study
to uncover this difference in adults. This finding may be attributed to
the divergent pharmacokinetics of PFOA between genders. In a recent
study of PFOA exposure in rats, female rats excreted approximately
60% or more of a PFOA dose in urine and faeces within 24 h, whereas
male rats excreted only 9.0% of the same PFOA dose in urine within a
prolonged period of 12 days (Kim et al., 2016). Previous studies have
also suggested that gender difference for the effect of PFOA exposure
could be partially explained by the specific excretion pathway of fe-
males, such as menstruation, delivery and/or lactation (Li et al., 2017;
Y. Zhang et al., 2014). In addition, PFOA is one of the potential endocrine
disrupting chemicals (Kjeldsen and Bonefeld-Jorgensen, 2013) and it
can interact with estrogen receptor α (Buhrke et al., 2015; Kang et al.,
2016).
In our study, we examined the associations between the serum PFOA
levels and glucose metabolism markers, comprising FPG, Fins and
HOMA-IR in adults. Consistent with some previous studies (Domazet
et al., 2016; Fisher et al., 2013; Karnes et al., 2014; Lin et al., 2009;
Lind et al., 2014; MacNeil et al., 2009; Nelson et al., 2010), none of
these analysed parameters were related to PFOA exposure, although
small number of population was reported in these studies. Besides
these analysed parameters above, beta cell is central to the develop-
ment and progression of diabetes (Cnop et al., 2005; Halban et al.,
2014). Previous studies have suggested the relationship between
PFOA and beta cell function. In a multicenter prospective investigation,
childhood PFOA exposure could predict impaired beta cell function at
15 years of age (Domazet et al., 2016). However, in a NHANES study ex-
ploring the role of PFOA/PFOS isomers in glucose homeostasis, increased
linear PFOA was associated with an enhancement of beta cell function
(Liu et al., 2017). In addition, strong positive association between
serum PFOA concentrations and β-cell function was observed in 969
adults from the NHANES 1999–2004 (Lin et al., 2009). Thus, the exact
Females
P for trend % diff 95% CI P for trend
b0.001 Reference 0.001
0.53% −0.18%, 1.25%
1.16%a 0.44%, 1.97%
1.07%a 0.27%, 1.97%
 X. He et al. / Science of the Total Environment 625 (2018) 566–574 571

mechanism that links PFOA to glucose metabolism is still unknown and
mixed.
In contrast, the evidence for a relationship between PFOA exposure
and lipid metabolism is much stronger. Although results in humans
are not entirely consistent, the general trend is one of positive associa-
tions between PFOA concentration and TC levels, mainly in occupational
workers (Costa et al., 2009; Sakr et al., 2007) or in population of com-
munities surrounding the DuPont chemical plant (Steenland et al.,
2009). Notably, we found that women in Q3 of PFOA levels showed a
greater increase in serum TC than those participants in Q4 of PFOA
levels. Thus, one should be cautious about concluding that PFOA may
have a relationship with TC in women. But interestingly, the significant
association was only observed in women aged over 70. This age differ-
ence may be the reason for the disappearance of PFOA dose response
in women. The aforementioned prospective study conducted in Danish
individuals found an association between PFOA and high TC levels in a
subset of participants 50–60 years old (Eriksen et al., 2013). However,
its population was limited to 50–60 years old and difference between
genders was not analysed. Two previous studies using NHANES data
also reported a relationship between serum PFOA and lipid metabolism
among general US adults (Liu et al., 2017; Nelson et al., 2010). However,
the results from these two study were different from ours. Specifically,
Nelson et al. found that associations were identical in each age and
sex subgroups with greater magnitude among persons 60–80 years of
age. In the study conducted by Liu et al., although increased linear
PFOA was associated with increases in TC, age and sex subgroups
were not implemented. In a study conducted in post-weaning rats,
age effect on PFOA plasma concentration has been investigated. A
slightly but significantly higher plasma concentration occurred in
older males while significantly lower (approximately 10-fold) plasma
PFOA concentrations was observed in older females (Hinderliter et al.,
2006). More epidemiological and mechanistic studies are needed to
verify the age difference of PFOA exposure effect and to elucidate the
potential reasons.
The underlying molecular mechanism for the association between
PFOA and diabetes has yet to be elucidated. It has been reported that
PFAS can bind to peroxisome proliferators activated receptors (PPARs)
(Lau et al., 2007). PPARα is mainly expressed in the liver and is involved
in lipid homeostasis, fatty acid catabolism and inflammation (Wolf et al.,
2008). PPARγ activation has been shown to influence insulin resistance
and insulin secretion (Lebovitz and Banerji, 2001; Lupi et al., 2004). The
fibrate class of cholesterol-lowering medications (one of the PPARα li-
gands) may inhibit secretion of cholesterol from the liver, resulting in
reduced cholesterol in the serum (Kennedy et al., 2004). However, our
results are not completely consistent with prior animal findings, sug-
gesting PFOA might exert its effects via alternative or even multiple
pathways. Furthermore, interspecies differences may partially explain
the inconsistency of cholesterol findings between animal and human
studies.
A recent epidemiological study conducted in a Chinese adult male
cohort has suggested that low-level environmental exposure to PFAS
was related to oxidative stress in humans and increased the risk of dia-
betes (Wang et al., 2017). An in vitro study by Suh et al. discovered that
PFOA exposure was linked to increased oxidative stress and mitochon-
drial dysfunction, inducing apoptosis in pancreatic β-cells (Suh et al.,
2017). Additionally, PFOA can cause metabolome disorders involving
pollutant detoxification, antioxidation and nitric oxide signalling path-
ways. Furthermore, Yan et al. found that the serine/threonine protein
kinase (AKT) signalling pathway was activated in PFOA exposure (Yan
et al., 2015). These results indicate that PPAR-independent mechanisms
may exist for PFOA-related diabetogenesis and further efforts should be
made to establish the underlying molecular mechanism.

Table 5
Estimated percent difference (% diff) and 95% confidence intervals (95% CI) in serum total cholesterol (TC) concentrations in US adults 2003–2012 for each interquartile ratio (IQ ratio)
increase in serum PFOA levels stratified by age.

Age (years) TC Males Females

% diff 95% CI P for trend % diff 95% CI P for trend

20–30 Q1 Reference 0.111 Reference 0.226

Q2 0.62% −1.41%, 2.70% 1.88% 0.44%, 3.34%
Q3 −0.18% −2.10%, 1.79% 1.61% 0.00%, 3.25%
Q4 1.61% −0.27%, 3.62% −0.18% −2.20%, 1.88%
30–40 Q1 Reference 0.438 Reference 0.190
Q2 0.71% −1.32%, 2.70% −0.09% −1.67%, 1.43%
Q3 0.71% −1.15%, 2.60% 1.97% 0.27%, 3.80%
Q4 0.98% −0.88%, 2.88% 1.16% −1.06%, 3.43%
40–50 Q1 Reference 0.004 Reference 0.001
Q2 1.52% −0.62%, 3.71% 2.43% 0.89%, 4.08%
Q3 1.16% −0.88%, 3.25% 3.25% 1.43%, 5.00%
Q4 2.88%a 0.89%, 4.91% 1.43% −0.44%, 3.43%
50–60 Q1 Reference 0.002 Reference 0.610
Q2 0.53% −1.59%, 2.79% 0.71% −1.15%, 2.70%
Q3 2.24% 0.09%, 4.45% 0.80% −1.24%, 2.88%
Q4 2.52%a 0.44%, 4.73% 2.88% 0.80%, 5.00%
60–70 Q1 Reference 0.341 Reference 0.006
Q2 0.44% −1.67%, 2.61% −0.35% −2.37%, 1.79%
Q3 0.62% −1.59%, 2.79% 0.80% −1.24%, 2.88%
Q4 1.25% −0.88%, 3.34% 2.88%a 0.80%, 5.00%
70–80 Q1 Reference 0.453 Reference 0.965
Q2 −0.88% −3.06%, 1.43% −1.15% −3.32%, 1.16%
Q3 0.98% −1.32%, 3.34% −0.27% −2.46%, 1.97%
Q4 0.36% −1.85%, 2.61% 0.36% −1.76%, 2.43%
N80 Q1 Reference 0.466 Reference 0.591
Q2 1.34% −4.68%, 7.84% −4.51% −11.14%, 2.61%
Q3 −2.63% −8.33%, 3.34% 0.00% −7.35%, 7.84%
Q4 −1.24% −6.61%, 4.45% −1.41% −8.74%, 6.60%

TC was Log-transformed in model.

% diff = [(IQ ratiobeta) − 1] ∗ 100.
IQ ratio = 75th/25th percentiles of serum PFOA: 2.43.
Quartiles (Q) of PFOA (ng/ml): Q1: b2.1; Q2: 2.1–3.34; Q3: 3.34–5.1; Q4: N5.1.
PFOA, perfluorooctanoic acid.
a
P b 0.05.
572 X. He et al. / Science of the Total Environment 625 (2018) 566–574
PFOA has been determined worldwide in human blood. The
median concentration of PFOA in adult blood from Queensland,
Australia (6.4 ng/ml) (Toms et al., 2009) and Shenyang, China
(6.19 ng/ml) (Bao et al., 2017) were higher than those reported for
adults from other regions (b 0.5–1.61 ng/ml) (i.e., Korea, arctic
Russia, Uzbekistan and Afghanistan) (Cho et al., 2015; Hanssen
et al., 2013; Hemat et al., 2010). It should be noted that the exposure
of PFOA tended to deviate from the normal distribution. In the stud-
ies of Rocca et al. conducted in Italy, a dispersed distribution of PFOA
was observed, with a large fraction of samples at or below LOD and
part of samples at high levels (La Rocca et al., 2015; La Rocca et al.,
2014). Even though the background concentration of PFOA is lower
now than it was a decade ago, its long half-life could suggest the ef-
fect of bioaccumulation, leading to long-term body burdens and
health risks (Poet et al., 2016). Humans can be exposed to PFOA
through various routes, such as diet, drinking water, inhalation and
ingestion of indoor dust (Ferguson et al., 2013). Other than plasma
and tissues, PFOA can also exist in umbilical cord blood (Arbuckle
et al., 2013) and breast milk (Fromme et al., 2010). Thus, further ef-
forts are still needed to limit emissions of PFOA.
This study has some critical strengths. While many previous investi-
gations of PFOA health effects were based on small samples or high ex-
posure levels, the current study examined a relatively large sample size
with a general PFOA exposure level. In addition to the common con-
founders, we analysed other covariates that may have effects on the de-
velopment of diabetes, such as PIR, education level, energy intake, time
spent in front of a screen and serum cotinine. Furthermore, we carried
Fig. 3. Scatter plots and fitted lines with 95% confidence intervals (95% CI) for the relationship between PFOA concentrations and serum TC, for males aged 40–60 (A) and females aged
60–70 (B).
out separate analyses stratified by age and sex, which can explore the
sensitivity of different groups to these chemicals. Our study has several
limitations. We cannot rule out the possibility of reverse causation be-
cause of the cross-sectional study. In general, PFAS can be excreted
through the kidneys. The lower PFOA filtration rate in kidney will lead
to higher serum concentrations of PFAS. A Previous study has shown
that PFAS filtration rate was lower in diabetic patients than in normal
subjects (Conway et al., 2016). As such, it is possible that some of
these associations are due to reduced PFAS excretion among patients
with diabetes. Moreover, a lot of environmental chemicals (such as
polycyclic aromatic hydrocarbons (Alshaarawy et al., 2014), bisphenol
A (Ahmadkhaniha et al., 2014), phthalates (Mamtani et al., 2016), mer-
cury (He et al., 2013), pesticide (Saldana et al., 2007) and arsenic (Rhee
et al., 2013)) were potentially associated with the risk of diabetes. These
chemicals were not evaluated in our analysis, which may have an im-
pact on the association between PFOA and diabetes. Future studies
may need to assess interaction effect of different chemicals on the risk
of diabetes.
5. Conclusion
In conclusion, serum PFOA was found to be positively associated
with diabetes mellitus in men, independent of several covariates and
in accordance with emerging toxicology data. In particular, our findings
indicate that PFOA may disrupt cholesterol metabolism at environmen-
tally relevant exposures. Considering the limitations of this study,
 X. He et al. / Science of the Total Environment 625 (2018) 566–574
further prospective studies are needed to test the causality of this asso-
ciation and elucidate the intermediate pathways involved.
Acknowledgement
This study was supported by grants from the National Natural Sci-
ence Foundation of China (81370920, 81770773), Natural Science Foun-
dation of Jiangsu Province (BK20171499), Project of "Six Talents Peak of
Jiangsu Province (2013WSN-023)", Jiangsu Province's Key Medical Tal-
ents (co-construction) Program, Project of “333” Talent in Jiangsu Prov-
ince, Jiangsu Province Official Hospital Scientific Research Initial
Funding (RPF201501), and Jiangsu Province Official Hospital Talents
Construction Fund Research Project (IR2015101).
Conflicts of interest
The authors declare that there is no conflict of interest.
Appendix A. Supplementary data
Supplementary data to this article can be found online at https://doi.
org/10.1016/j.scitotenv.2017.12.186.
References
Ahmadkhaniha, R., Mansouri, M., Yunesian, M., Omidfar, K., Jeddi, M.Z., Larijani, B., et al.,
2014. Association of urinary bisphenol a concentration with type-2 diabetes mellitus.
J. Environ. Health Sci. Eng. 12, 64.
Alshaarawy, O., Zhu, M., Ducatman, A.M., Conway, B., Andrew, M.E., 2014. Urinary polycy-
clic aromatic hydrocarbon biomarkers and diabetes mellitus. Occup. Environ. Med.
71, 437–441.
Arbuckle, T.E., Kubwabo, C., Walker, M., Davis, K., Lalonde, K., Kosarac, I., et al., 2013. Um-
bilical cord blood levels of perfluoroalkyl acids and polybrominated flame retardants.
Int. J. Hyg. Environ. Health 216, 184–194.
Audouze, K., Brunak, S., Grandjean, P., 2013. A computational approach to chemical etiol-
ogies of diabetes. Sci. Rep. 3, 2712.
Bainbridge, K.E., Hoffman, H.J., Cowie, C.C., 2008. Diabetes and hearing impairment in the
United States: audiometric evidence from the National Health and Nutrition Exami-
nation Survey, 1999 to 2004. Ann. Intern. Med. 149, 1–10.
Bao, W.W., Qian, Z.M., Geiger, S.D., Liu, E., Liu, Y., Wang, S.Q., et al., 2017. Gender-specific
associations between serum isomers of perfluoroalkyl substances and blood pressure
among Chinese: isomers of C8 Health Project in China. Sci. Total Environ. 607-608,
1304–1312.
Bond, G.G., Dietrich, D.R., 2017. Further thoughts on limitations, uncertainties and com-
peting interpretations regarding chemical exposures and diabetes. J. Epidemiol. Com-
munity Health 71, 943.
Buck, R.C., Franklin, J., Berger, U., Conder, J.M., Cousins, I.T., de Voogt, P., et al., 2011.
Perfluoroalkyl and polyfluoroalkyl substances in the environment: terminology, clas-
sification, and origins. Integr. Environ. Assess. Manag. 7, 513–541.
Buhrke, T., Kruger, E., Pevny, S., Rossler, M., Bitter, K., Lampen, A., 2015. Perfluorooctanoic
acid (PFOA) affects distinct molecular signalling pathways in human primary hepato-
cytes. Toxicology 333, 53–62.
Butt, C.M., Berger, U., Bossi, R., Tomy, G.T., 2010. Levels and trends of poly- and
perfluorinated compounds in the arctic environment. Sci. Total Environ. 408,
2936–2965.
Calafat, A.M., Wong, L.Y., Kuklenyik, Z., Reidy, J.A., Needham, L.L., 2007. Polyfluoroalkyl
chemicals in the U.S. population: data from the National Health and Nutrition Exam-
ination Survey (NHANES) 2003–2004 and comparisons with NHANES 1999–2000.
Environ. Health Perspect. 115, 1596–1602.
Cho, C.R., Lam, N.H., Cho, B.M., Kannan, K., Cho, H.S., 2015. Concentration and correlations
of perfluoroalkyl substances in whole blood among subjects from three different geo-
graphical areas in Korea. Sci. Total Environ. 512-513, 397–405.
Christensen, K.Y., Raymond, M., Thompson, B.A., Anderson, H.A., 2016. Perfluoroalkyl sub-
stances in older male anglers in Wisconsin. Environ. Int. 91, 312–318.
Cnop, M., Welsh, N., Jonas, J.C., Jorns, A., Lenzen, S., Eizirik, D.L., 2005. Mechanisms of pan-
creatic beta-cell death in type 1 and type 2 diabetes: many differences, few similari-
ties. Diabetes 54 (Suppl. 2), S97–107.
Conway, B., Innes, K.E., Long, D., 2016. Perfluoroalkyl substances and beta cell deficient di-
abetes. J. Diabetes Complicat. 30, 993–998.
Costa, G., Sartori, S., Consonni, D., 2009. Thirty years of medical surveillance in
perfluooctanoic acid production workers. J. Occup. Environ. Med. 51, 364–372.
Domazet, S.L., Grontved, A., Timmermann, A.G., Nielsen, F., Jensen, T.K., 2016. Longitudinal
associations of exposure to perfluoroalkylated substances in childhood and adoles-
cence and indicators of adiposity and glucose metabolism 6 and 12 years later: the
European Youth Heart Study. Diabetes Care 39, 1745–1751.
Eriksen, K.T., Raaschou-Nielsen, O., McLaughlin, J.K., Lipworth, L., Tjonneland, A., Overvad,
K., et al., 2013. Association between plasma PFOA and PFOS levels and total cholester-
ol in a middle-aged Danish population. PLoS One 8, e56969.
573
Ferguson, K.K., O'Neill, M.S., Meeker, J.D., 2013. Environmental contaminant exposures
and preterm birth: a comprehensive review. J. Toxicol. Environ. Health B Crit. Rev.
16, 69–113.
Fisher, M., Arbuckle, T.E., Wade, M., Haines, D.A., 2013. Do perfluoroalkyl substances affect
metabolic function and plasma lipids?—analysis of the 2007–2009, Canadian Health
Measures Survey (CHMS) Cycle 1. Environ. Res. 121, 95–103.
Fromme, H., Mosch, C., Morovitz, M., Alba-Alejandre, I., Boehmer, S., Kiranoglu, M., et al.,
2010. Pre- and postnatal exposure to perfluorinated compounds (PFCs). Environ.
Sci. Technol. 44, 7123–7129.
Goldfield, G.S., Saunders, T.J., Kenny, G.P., Hadjiyannakis, S., Phillips, P., Alberga, A.S., et al.,
2013. Screen viewing and diabetes risk factors in overweight and obese adolescents.
Am. J. Prev. Med. 44, S364–370.
Gore, A.C., Chappell, V.A., Fenton, S.E., Flaws, J.A., Nadal, A., Prins, G.S., et al., 2015. EDC-2:
the Endocrine Society's second scientific statement on endocrine-disrupting
chemicals. Endocr. Rev. 36, E1–E150.
Halban, P.A., Polonsky, K.S., Bowden, D.W., Hawkins, M.A., Ling, C., Mather, K.J., et al., 2014.
β-Cell failure in type 2 diabetes: postulated mechanisms and prospects for preven-
tion and treatment. J. Clin. Endocrinol. Metab. 99, 1983–1992.
Hanssen, L., Dudarev, A.A., Huber, S., Odland, J.O., Nieboer, E., Sandanger, T.M., 2013. Par-
tition of perfluoroalkyl substances (PFASs) in whole blood and plasma, assessed in
maternal and umbilical cord samples from inhabitants of arctic Russia and
Uzbekistan. Sci. Total Environ. 447, 430–437.
He, K., Xun, P., Liu, K., Morris, S., Reis, J., Guallar, E., 2013. Mercury exposure in young
adulthood and incidence of diabetes later in life: the CARDIA Trace Element Study. Di-
abetes Care 36, 1584–1589.
Hemat, H., Wilhelm, M., Volkel, W., Mosch, C., Fromme, H., Wittsiepe, J., 2010. Low serum
levels of perfluorooctanoic acid (PFOA), perfluorooctane sulfonate (PFOS) and
perfluorohexane sulfonate (PFHxS) in children and adults from Afghanistan. Sci.
Total Environ. 408, 3493–3495.
Hinderliter, P.M., Han, X., Kennedy, G.L., Butenhoff, J.L., 2006. Age effect on
perfluorooctanoate (PFOA) plasma concentration in post-weaning rats following
oral gavage with ammonium perfluorooctanoate (APFO). Toxicology 225, 195–203.
Holst, C., Becker, U., Jorgensen, M.E., Gronbaek, M., Tolstrup, J.S., 2017. Alcohol drinking
patterns and risk of diabetes: a cohort study of 70,551 men and women from the gen-
eral Danish population. Diabetologia 60, 1941–1950.
Honda, T., Pun, V.C., Manjourides, J., Suh, H., 2017. Associations between long-term expo-
sure to air pollution, glycosylated hemoglobin and diabetes. Int. J. Hyg. Environ.
Health 220, 1124–1132.
IDF Diabetes Atlas Group, 2015. Update of mortality attributable to diabetes for the IDF
Diabetes Atlas: estimates for the year 2013. Diabetes Res. Clin. Pract. 109, 461–465.
Jain, R.B., 2014. Contribution of diet and other factors to the levels of selected polyfluorinated
compounds: data from NHANES 2003–2008. Int. J. Hyg. Environ. Health 217, 52–61.
James-Todd, T., Stahlhut, R., Meeker, J.D., Powell, S.G., Hauser, R., Huang, T., et al., 2012.
Urinary phthalate metabolite concentrations and diabetes among women in the Na-
tional Health and Nutrition Examination Survey (NHANES) 2001–2008. Environ.
Health Perspect. 120, 1307–1313.
Jarvandi, S., Schootman, M., Racette, S.B., 2015. Breakfast intake among adults with type 2
diabetes: influence on daily energy intake. Public Health Nutr. 18, 2146–2152.
Kang, J.S., Choi, J.S., Park, J.W., 2016. Transcriptional changes in steroidogenesis by
perfluoroalkyl acids (PFOA and PFOS) regulate the synthesis of sex hormones in
H295R cells. Chemosphere 155, 436–443.
Karnes, C., Winquist, A., Steenland, K., 2014. Incidence of type II diabetes in a cohort with
substantial exposure to perfluorooctanoic acid. Environ. Res. 128, 78–83.
Kennedy Jr., G.L., Butenhoff, J.L., Olsen, G.W., O'Connor, J.C., Seacat, A.M., Perkins, R.G., et
al., 2004. The toxicology of perfluorooctanoate. Crit. Rev. Toxicol. 34, 351–384.
Kim, S.J., Heo, S.H., Lee, D.S., Hwang, I.G., Lee, Y.B., Cho, H.Y., 2016. Gender differences in
pharmacokinetics and tissue distribution of 3 perfluoroalkyl and polyfluoroalkyl sub-
stances in rats. Food Chem. Toxicol. 97, 243–255.
Kjeldsen, L.S., Bonefeld-Jorgensen, E.C., 2013. Perfluorinated compounds affect the func-
tion of sex hormone receptors. Environ. Sci. Pollut. Res. Int. 20, 8031–8044.
La Rocca, C., Tait, S., Guerranti, C., Busani, L., Ciardo, F., Bergamasco, B., et al., 2014. Exposure
to endocrine disrupters and nuclear receptor gene expression in infertile and fertile
women from different Italian areas. Int. J. Environ. Res. Public Health 11, 10146–10164.
La Rocca, C., Tait, S., Guerranti, C., Busani, L., Ciardo, F., Bergamasco, B., et al., 2015. Expo-
sure to endocrine disruptors and nuclear receptors gene expression in infertile and
fertile men from Italian areas with different environmental features. Int. J. Environ.
Res. Public Health 12, 12426–12445.
Lau, C., Anitole, K., Hodes, C., Lai, D., Pfahles-Hutchens, A., Seed, J., 2007. Perfluoroalkyl
acids: a review of monitoring and toxicological findings. Toxicol. Sci. 99, 366–394.
Lebovitz, H.E., Banerji, M.A., 2001. Insulin resistance and its treatment by thiazolidinediones.
Recent Prog. Horm. Res. 56, 265–294.
Leonard, R.C., Kreckmann, K.H., Sakr, C.J., Symons, J.M., 2008. Retrospective cohort mortal-
ity study of workers in a polymer production plant including a reference population
of regional workers. Ann. Epidemiol. 18, 15–22.
Li, Y., Cheng, Y., Xie, Z., Zeng, F., 2017. Perfluorinated alkyl substances in serum of the
southern Chinese general population and potential impact on thyroid hormones.
Sci. Rep. 7, 43380.
Lin, C.Y., Chen, P.C., Lin, Y.C., Lin, L.Y., 2009. Association among serum perfluoroalkyl
chemicals, glucose homeostasis, and metabolic syndrome in adolescents and adults.
Diabetes Care 32, 702–707.
Lind, L., Zethelius, B., Salihovic, S., van Bavel, B., Lind, P.M., 2014. Circulating levels of
perfluoroalkyl substances and prevalent diabetes in the elderly. Diabetologia 57,
473–479.
Liu, H.S., Wen, L.L., Chu, P.L., Lin, C.Y., 2017. Association among total serum isomers of
perfluorinated chemicals, glucose homeostasis, lipid profiles, serum protein and met-
abolic syndrome in adults: NHANES, 2013–2014. Environ. Pollut. 232, 73–79.
574 X. He et al. / Science of the Total Environment 625 (2018) 566–574
Lorenzo, C., Williams, K., Hunt, K.J., Haffner, S.M., 2007. The National Cholesterol Educa-
tion Program - Adult Treatment Panel III, International Diabetes Federation, and
World Health Organization definitions of the metabolic syndrome as predictors of in-
cident cardiovascular disease and diabetes. Diabetes Care 30, 8–13.
Lupi, R., Del Guerra, S., Marselli, L., Bugliani, M., Boggi, U., Mosca, F., et al., 2004.
Rosiglitazone prevents the impairment of human islet function induced by fatty
acids: evidence for a role of PPARgamma2 in the modulation of insulin secretion.
Am. J. Physiol. Endocrinol. Metab. 286, E560–567.
MacNeil, J., Steenland, N.K., Shankar, A., Ducatman, A., 2009. A cross-sectional analysis of
type II diabetes in a community with exposure to perfluorooctanoic acid (PFOA). En-
viron. Res. 109, 997–1003.
Mamtani, M., Curran, J.E., Blangero, J., Kulkarni, H., 2016. Association of Urinary phthalates
with self-reported eye affliction/retinopathy in individuals with diabetes: National
Health and Nutrition Examination Survey, 2001–2010. J. Diabetes Res. 2016,
7269896.
Menke, A., Orchard, T.J., Imperatore, G., Bullard, K.M., Mayer-Davis, E., Cowie, C.C., 2013.
The prevalence of type 1 diabetes in the United States. Epidemiology 24, 773–774.
Moon, S.S., 2014. Additive effect of heavy metals on metabolic syndrome in the Korean
population: the Korea National Health and Nutrition Examination Survey
(KNHANES) 2009–2010. Endocrine 46, 263–271.
Moyer, J., Wilson, D., Finkelshtein, I., Wong, B., Potts, R., 2012. Correlation between sweat
glucose and blood glucose in subjects with diabetes. Diabetes Technol. Ther. 14,
398–402.
Nelson, J.W., Hatch, E.E., Webster, T.F., 2010. Exposure to polyfluoroalkyl chemicals and
cholesterol, body weight, and insulin resistance in the general U.S. population. Envi-
ron. Health Perspect. 118, 197–202.
Ng, C.A., Hungerbuehler, K., 2015. Exploring the use of molecular docking to identify
bioaccumulative perfluorinated alkyl acids (PFAAs). Environ. Sci. Technol. 49,
12306–12314.
Olsen, G.W., Burris, J.M., Ehresman, D.J., Froehlich, J.W., Seacat, A.M., Butenhoff, J.L., et al., 2007.
Half-life of serum elimination of perfluorooctanesulfonate,perfluorohexanesulfonate,
and perfluorooctanoate in retired fluorochemical production workers. Environ. Health
Perspect. 115, 1298–1305.
Peng, Q., Harlow, S.D., Park, S.K., 2015. Urinary arsenic and insulin resistance in US adoles-
cents. Int. J. Hyg. Environ. Health 218, 407–413.
Poet, T.S., Schlosser, P.M., Rodriguez, C.E., Parod, R.J., Rodwell, D.E., Kirman, C.R., 2016.
Using physiologically based pharmacokinetic modeling and benchmark dose
methods to derive an occupational exposure limit for N-methylpyrrolidone. Regul.
Toxicol. Pharmacol. 76, 102–112.
Post, G.B., Cohn, P.D., Cooper, K.R., 2012. Perfluorooctanoic acid (PFOA), an emerging
drinking water contaminant: a critical review of recent literature. Environ. Res. 116,
93–117.
Rhee, S.Y., Hwang, Y.C., Woo, J.T., Chin, S.O., Chon, S., Kim, Y.S., 2013. Arsenic exposure and
prevalence of diabetes mellitus in Korean adults. J. Korean Med. Sci. 28, 861–868.
Sakr, C.J., Kreckmann, K.H., Green, J.W., Gillies, P.J., Reynolds, J.L., Leonard, R.C., 2007.
Cross-sectional study of lipids and liver enzymes related to a serum biomarker of ex-
posure (ammonium perfluorooctanoate or APFO) as part of a general health survey in
a cohort of occupationally exposed workers. J. Occup. Environ. Med. 49, 1086–1096.
Saldana, T.M., Basso, O., Hoppin, J.A., Baird, D.D., Knott, C., Blair, A., et al., 2007. Pesticide
exposure and self-reported gestational diabetes mellitus in the Agricultural Health
Study. Diabetes Care 30, 529–534.
Steenland, K., Woskie, S., 2012. Cohort mortality study of workers exposed to
perfluorooctanoic acid. Am. J. Epidemiol. 176, 909–917.
Steenland, K., Tinker, S., Frisbee, S., Ducatman, A., Vaccarino, V., 2009. Association of
perfluorooctanic acid (PFOA) and perfluoroctanesulfonate (PFOS) with serum lipids
among adults living near a chemical plant. 20.
Su, T.C., Kuo, C.C., Hwang, J.J., Lien, G.W., Chen, M.F., Chen, P.C., 2016. Serum
perfluorinated chemicals, glucose homeostasis and the risk of diabetes in working-
aged Taiwanese adults. Environ. Int. 88, 15–22.
Suh, K.S., Choi, E.M., Kim, Y.J., Hong, S.M., Park, S.Y., Rhee, S.Y., et al., 2017.
Perfluorooctanoic acid induces oxidative damage and mitochondrial dysfunction in
pancreatic beta-cells. Mol. Med. Rep. 15, 3871–3878.
Thiering, E., Bruske, I., Kratzsch, J., Thiery, J., Sausenthaler, S., Meisinger, C., et al., 2011.
Prenatal and postnatal tobacco smoke exposure and development of insulin resis-
tance in 10 year old children. Int. J. Hyg. Environ. Health 214, 361–368.
Toms, L.M., Calafat, A.M., Kato, K., Thompson, J., Harden, F., Hobson, P., et al., 2009.
Polyfluoroalkyl chemicals in pooled blood serum from infants, children, and adults
in Australia. Environ. Sci. Technol. 43, 4194–4199.
Wang, Z., Cousins, I.T., Scheringer, M., Hungerbuhler, K., 2013. Fluorinated alternatives to
long-chain perfluoroalkyl carboxylic acids (PFCAs), perfluoroalkane sulfonic acids
(PFSAs) and their potential precursors. Environ. Int. 60, 242–248.
Wang, X., Liu, L., Zhang, W., Zhang, J., Du, X., Huang, Q., et al., 2017. Serum metabolome
biomarkers associate low-level environmental perfluorinated compound exposure
with oxidative/nitrosative stress in humans. Environ. Pollut. 229, 168–176.
Wolf, C.J., Takacs, M.L., Schmid, J.E., Lau, C., Abbott, B.D., 2008. Activation of mouse and
human peroxisome proliferator-activated receptor alpha by perfluoroalkyl acids of
different functional groups and chain lengths. Toxicol. Sci. 106, 162–171.
Xu, C., Liu, Q., Zhang, Q., Gu, A., Jiang, Z.Y., 2015. Urinary enterolactone is associated with
obesity and metabolic alteration in men in the US National Health and Nutrition Ex-
amination Survey 2001–10. Br. J. Nutr. 113, 683–690.
Yan, S., Zhang, H., Zheng, F., Sheng, N., Guo, X., Dai, J., 2015. Perfluorooctanoic acid expo-
sure for 28 days affects glucose homeostasis and induces insulin hypersensitivity in
mice. Sci. Rep. 5, 11029.
Zhang, Y., Jiang, W., Fang, S., Zhu, L., Deng, J., 2014. Perfluoroalkyl acids and the isomers of
perfluorooctanesulfonate and perfluorooctanoate in the sera of 50 new couples in
Tianjin, China. Environ. Int. 68, 185–191.
Zhang, Z., Kris-Etherton, P.M., Hartman, T.J., 2014. Birth weight and risk factors for cardio-
vascular disease and type 2 diabetes in US children and adolescents: 10 year results
from NHANES. Matern. Child Health J. 18, 1423–1432.
Zhang, C., Sundaram, R., Maisog, J., Calafat, A.M., Barr, D.B., Buck Louis, G.M., 2015. A pro-
spective study of prepregnancy serum concentrations of perfluorochemicals and the
risk of gestational diabetes. Fertil. Steril. 103, 184–189.