Environment International 156 (2021) 106599

Contents lists available at ScienceDirect

Environment International
journal homepage: www.elsevier.com/locate/envint

Perfluoroalkyl substances and immune cell counts in adults from the

Mid-Ohio Valley (USA)
Maria-Jose Lopez-Espinosa a, b, c, *, Christian Carrizosa a, Michael I. Luster d, Joseph
B. Margolick e, Olga Costa a, Giovanni S. Leonardi f, g, Tony Fletcher f, g
a
Epidemiology and Environmental Health Joint Research Unit, FISABIO− Universitat Jaume I− Universitat de València, 46020 Valencia, Spain
b
Spanish Consortium for Research on Epidemiology and Public Health (CIBERESP), 28029 Madrid, Spain
c
Faculty of Nursing and Chiropody, Universitat de València, 46001 Valencia, Spain
d
School of Public Health, West Virginia University, Morgantown, WV 26505, USA
e
Johns Hopkins Bloomberg School of Public Health, Johns Hopkins University, Baltimore, MD 21279, USA
f
Centre for Radiation, Chemical and Environmental Hazards, Public Health England (PHE), Chilton, Oxfordshire OX11 0RQ, United Kingdom
g
Department of Public Health, Environments and Society, London School of Hygiene and Tropical Medicine, London WCIH 9SH, United Kingdom

A R T I C L E I N F O A B S T R A C T

Handling Editor: Shoji F. Nakayama Background: Although perfluoroalkyl substances (PFASs) may be immunotoxic, evidence for this in humans is
scarce. We studied the association between 4 PFASs (perfluorohexane sulfonate [PFHxS], perfluorooctanoic acid
Keywords: [PFOA], perfluorooctane sulfonate [PFOS] and perfluorononanoic acid [PFNA]) and circulating levels of several
PFASs types of immune cells.
Immune system
Methods: Serum PFASs and white blood cell types were measured in 42,782 (2005–2006) and 526 (2010) adults
White blood cells
from an area with PFOA drinking water contamination in the Mid-Ohio Valley (USA). Additionally, the major
Lymphocytes
lymphocyte subsets were measured in 2010. Ln(cell counts) and percentages of cell counts were regressed on
serum PFAS concentrations (ln or percentiles). Adjusted results were expressed as the percentage difference (95%
CI) per interquartile range (IQR) increment of each PFAS concentration.
Results: Generally positive monotonic associations between total lymphocytes and PFHxS, PFOA, and PFOS were
found in both surveys (difference range: 1.12–7.33% for count and 0.36–1.77 for percentage, per PFAS IQR
increment), and were stronger for PFHxS. These associations were reflected in lymphocyte subset counts but not
percentages, with PFHxS positively and monotonically associated with T, B, and natural killer (NK) cell counts
(range: 5.51–8.62%), PFOA and PFOS with some T-cell phenotypes, and PFOS with NK cells (range:
3.12–12.21%), the associations being monotonic in some cases. Neutrophils, particularly percentage (range:
− 1.74 to − 0.36), showed decreasing trends associated with PFASs. Findings were less consistent for monocytes
and eosinophils.
Conclusion: These results suggest an association between PFHxS and, less consistently, for PFOA and PFOS, and
total lymphocytes (although the magnitudes of the differences were small). The increase in absolute lymphocyte
count appeared to be evenly distributed across lymphocyte subsets since associations with their percentages were
not significant.

1. Introduction
Perfluoroalkyl substances (PFASs) are a diverse group of chemicals
used in industrial applications and consumer products since the 1950s.
They persist in the environment, resist degradation, bioaccumulate in
humans, and have relatively long half-lives in the body of humans
(approximately 4–6 years) (Agency for Toxic Substances and Disease
Registry [ATSDR], 2018). Whereas mean serum concentrations of PFASs
have decreased in recent years in Western countries (Sunderland et al.,
2019), probably because their production has been reduced or has been
banned (ATSDR, 2018), the stability of these compounds and recent
increases in serum levels observed in Asia and the Middle East

* Corresponding author at: Foundation for the Promotion of Health and Biomedical Research in the Valencian Region, FISABIO-Public Health, Avenida de Cat­
alunya 21, 46020 Valencia, Spain.
E-mail address: lopez_josesp@gva.es (M.-J. Lopez-Espinosa).

https://doi.org/10.1016/j.envint.2021.106599
Received 22 December 2020; Received in revised form 18 March 2021; Accepted 21 April 2021
Available online 13 May 2021
0160-4120/© 2021 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
M.-J. Lopez-Espinosa et al.
(International Pollutants Elimination Network, 2019) mean that PFASs,
and their possible health effects, remain significant public health
concerns.
To date, most of the animal and human studies on the health effects
of some PFASs have focused on perfluorooctanoic acid (PFOA) and
perfluorooctane sulfonate (PFOS), and to a lesser extent on per­
fluorohexane sulfonate (PFHxS) and perfluorononanoic acid (PFNA).
The results of those studies suggest that they may affect metabolic,
endocrine, and immune processes (Ballesteros et al., 2017; EFSA [Eu­
ropean Food Safety Authority], 2018; Lau et al., 2007; Matilla-
Santander et al., 2017; United States Department of Health and
Human Services-National Toxicology Program [USDHHS-NTP], 2016).
Regarding the immune system, an extensive review of the literature by
the United States National Toxicology Program (USDHHS-NTP, 2016)
concluded that both PFOA and PFOS are immunotoxic. This conclusion
was based primarily upon a high level of evidence from animal studies
showing that at high concentrations they suppress the antibody
response, and a moderate level of evidence from human studies indi­
cating that they affect vaccine responses. This conclusion is also sup­
ported by a more recent review from the European Food Safety
Authority (EFSA), which reported an adverse effect on serum antibody
response after vaccination due to PFOS (and possibly PFOA) exposure,
with children considered to be the most vulnerable group (EFSA, 2018).
Regarding PFHxS and PFNA exposure, there is still insufficient epide­
miological data on their possible immunotoxicity (Chen et al., 2018;
Dong et al., 2013; Grandjean et al., 2017) to reach a conclusion.
Most observational studies in humans have investigated associations
between PFASs and immune outcomes such as asthma, allergies, pro­
pensity to infections, and serum antibody response to vaccination
(EFSA, 2018). However, immunotoxicity can also be studied by
measuring peripheral counts of white blood cells (WBCs: neutrophils,
lymphocytes, monocytes, eosinophils, and basophils) and subsets of
peripheral blood lymphocytes delineated by cell-surface markers
(CD3+T cells, CD3+CD4+T helper cells, CD3+CD8+T-cytotoxic cells,
CD3+CD4+ CD8+ double positive [DP] T cells, CD3− CD16+ CD56+
natural killer [NK] cells, CD3− CD19+B cells, and CD4+/CD8+ ratio).
WBC count and peripheral blood lymphocyte measurements are
powerful diagnostic tools used primarily in the identification and pro­
gression of diseases such as acquired immunodeficiency syndrome
(AIDS), and leukemias and lymphomas (Castilho et al., 2016; Davis
et al., 2014). They have also been used occasionally in clinical studies for
testing both the efficacy and the toxicity of therapeutics (e.g., Aziz et al.,
2013), and to help identify immunotoxicity following environmental
chemical exposure (Haase et al., 2016; Leonardi et al., 2000; Tryphonas,
2001). Nevertheless, data obtained from human epidemiologic studies
on PFAS exposure and these immune markers is still limited, with some
data derived from monitoring humans exposed to PFASs in the work­
place (Costa et al., 2009; Olsen et al., 2003) or from cross-sectional
(Abraham et al., 2020; Dong et al., 2013; Emmett et al., 2006; Gay­
lord et al., 2019; Knudsen et al., 2018) or longitudinal (Oulhote et al.,
2017) non-occupational human studies. Furthermore, only a limited
number of immunological end points have been studied in most of the
previous studies (predominantly, total WBCs, and/or eosinophils).
Given these limitations, further studies are warranted.
Previous work by our group showed that PFOA may be associated
with a decrease in influenza vaccine efficacy in adults who had had an
A/H3N2 seasonal flu vaccination (Looker et al., 2014). To further
address immunotoxicity, we aimed to evaluate the association between
PFAS concentrations and WBC counts (measured in both 2005–2006 and
2010), and the major lymphocyte subsets (measured in 2010) amongst
adults who had been highly exposed to PFOA due to contaminated water
supplies, but with similar exposure to other PFASs compared to other
populations from North America (Frisbee et al., 2009).
2
Environment International 156 (2021) 106599
2. Methods
2.1. Study design and subjects
The C8 Health Project arose from the settlement of a class action
lawsuit filed in 2002 by residents living near the DuPont Washington
Works plant in West Virginia regarding the release of PFOA from the
facility that had contaminated drinking water supplies along the Mid-
Ohio River Valley. The project collected data on 69,030 subjects (esti­
mated participation rate for residents at the time: 80%) between August
2005 and August 2006. Individuals were eligible for the study if they
had consumed water for at least one year between 1950 and 2004 from
any of the six contaminated water districts or from private wells in the
proximity of the chemical plant. Details of the study population have
been described elsewhere (Frisbee et al., 2009).
Between March and July 2010, participants were recalled and
invited to participate in a second survey, as described elsewhere (Fitz-
Simon et al., 2013). Briefly, the second survey was restricted to residents
of the three Ohio and five West Virginia ZIP code areas in the region
most affected by PFOA contamination. The follow-up population was
restricted to those aged 20–59 at the time of the initial survey
(2005–2006), and individuals were chosen randomly from age–gender
strata to give balanced numbers of men and women in each ten-year age
group. Altogether 973 people were contacted, and after excluding peo­
ple who reported a past or present diagnosis of cancer and those un­
willing to provide blood samples, 748 participated, with ages between
24 and 64 years at the time of the second survey. From the first survey
containing participants of all ages, we focused on the subset of adults
aged 24–64 years at baseline (2005–2006) so as to be in the same age
range as those in the follow-up study (2010).
Information on infections was not available in the first survey, but
this information was available in 2010. Participants were asked if, at the
time of the survey, they had any of the following infections or conditions
associated with inflammation: cold, influenza, sore throat, bronchitis,
pneumonia, gastroenteritis, cold sore, shingles, and/or sinus, ear, tooth,
and/or mouth infections. A total of 213 out of 748 participants reported
at least one of the above-mentioned conditions, with sinus and skin in­
fections being the most common. Analyses for the follow-up study
focused on those who reported not having any infections or conditions
associated with inflammation plus PFAS analyses. We chose to exclude
those with infections as they could have an effect on immune cell counts
(George-Gay and Parker, 2003). Altogether, 42,782 and 526 adults in
the first and second surveys, respectively, gave a blood sample, had
analyses of both PFASs and immune markers (2005–2006: types of
WBCs, and 2010: types of WBCs and subtypes of lymphocytes), as well as
data on sociodemographic and anthropometric factors, and were
included in the present study. The London School of Hygiene and
Tropical Medicine Ethics Committee approved this study and informed
consent was obtained from all participants in each period.
2.2. Serum PFAS concentrations
Laboratory analyses of serum PFASs were conducted by Exygen,
State College, PA, USA (a commercial laboratory) in 2005–2006 and by
the Centers for Disease Control and Prevention (CDC) in Atlanta, GA,
USA in 2010. The analytical (Flaherty et al., 2005; Frisbee et al., 2009;
Kato et al., 2011) and quality control (Van Leeuwen et al., 2006) pro­
cedures employed by both laboratories have been described elsewhere.
Briefly, laboratory analyses of PFASs used online solid phase extraction
coupled with reversed-phase high-performance liquid chromatography
separation and detection by isotope-dilution tandem mass spectrometry.
The limits of detection (LODs) in 2005–2006 were 0.5 ng/mL for all
PFASs. In 2010, LODs were 0.1 ng/mL for PFNA and PFHxS, 0.2 ng/mL
for PFOS, and 0.5 ng/mL for PFOA.
M.-J. Lopez-Espinosa et al.
2.3. Immune cell analysis
Whole blood collected in EDTA vacutainer tubes was shipped to
laboratories for analysis of immune cells. In LabCorp Inc., Burlington,
NC, USA (an accredited clinical diagnostic laboratory), analyses of types
of WBCs (neutrophils, monocytes, eosinophils, lymphocytes, and baso­
phils) were performed using an automated hematology analyzer (LH
Series, Beckman Coulter). In the flow cytometry core laboratory at Johns
Hopkins Bloomberg School of Public Health in Baltimore, immunophe­
notypic analysis was performed on peripheral blood lymphocytes
(CD3+ T cells, CD3+ CD4+ T helper cells, CD3+ CD8+ T-cytotoxic cells,
CD3+ CD4+ CD8+ DP T cells, CD3− CD16+ CD56+ NK cells, CD3−
CD19+ B cells) using standard flow cytometry methods, as previously
described (Giorgi et al., 1990; Schenker et al., 1993). The CD4+/CD8+
ratio was also calculated. The laboratory was certified by two profi­
ciency testing programs, i.e., the Immunology Quality Assurance (IQA)
administered by the US National Institutes of Health, and the College of
American Pathologists (CAP) program. The laboratory was also certified
under the provisions of the Clinical Laboratory Improvement Act (CLIA).
2.4. Covariates
We obtained information on sociodemographic and anthropometric
characteristics, as well as on lifestyle variables, through questionnaires
that were administered by trained interviewers in person (2005–2006)
and by telephone (2010). We considered the following variables:
gender, race/ethnicity, age, education, body mass index (BMI, only
available in the 2005–2006 survey), tobacco consumption, alcohol
intake, present or past diagnosis of immune diseases and/or cancer, and
anti-inflammatory medication (only available for the 2010 survey). The
participants were asked if they were taking or had taken in the last four
years paracetamol, aspirin, or other anti-inflammatory medications on a
regular basis (i.e., at least once a week for six months or more).
Serum creatinine levels were determined using the Kinetic Jaffe
method (LabCorp, Burlington, NC). We estimated glomerular filtration
rate (eGFR) using the Modification of Diet in Renal Disease (MDRD)
Study 4-variable equation (Levey et al., 2006).
2.5. Statistical analysis
We performed statistical analyses for all immune cell types except
basophils (due to the low proportion of samples with these immune cells
present: 25.6% in 2005–2006 and 17.9% in 2010). For PFASs, the %
≤LOD was very low (<0.1% for PFOA up to 2.6% for PFHxS) and they
were set to half the LOD for the analyses. Pearson’s correlations between
the four ln(PFASs), and these contaminants and eGFR were performed,
and a heatmap was also generated to display them. We used ordinal
mixed models to assess the baseline to follow-up differences in socio­
demographic characteristics. Linear mixed models adjusted for cova­
riates were also employed to study the changes in exposure levels of ln
(PFASs) over time.
We used multivariate linear regression analysis to assess the relation
between ln(counts) and percentages of immune cells and ln(PFASs) at
the baseline (n = 42,782) and follow-up (n = 526). All models were
adjusted for gender, age in 10-year categories, smoking, month of
sampling (because of a decreasing trend in serum PFASs during the
collection period in this population), alcohol intake, and educational
level (hereafter called ‘main analysis’).
We also conducted several sensitivity analyses including: BMI (since
excess adiposity has been related to both impaired immune function
(Martí et al., 2001) and PFASs (Braun, 2017; Cardenas et al., 2018), we
preferred not to include this variable in the main analysis); the four
PFASs simultaneously; eGFR (to evaluate whether a common disease
process may be contributing to altered PFAS excretion from the kidney
and increasing immune cell counts) (Dhingra et al., 2017; Watkins et al.,
2013); and use of anti-inflammatory drugs (only available in the 2010
3
Environment International 156 (2021) 106599
survey). We also excluded from the models those participants with self-
reported autoimmune diseases (n = 1,606 in 2005–2006, n = 28 in
2010) to account for medical conditions that could affect immunological
measurements (specifically, systemic lupus erythematosus, scleroderma,
multiple sclerosis, Sjögren’s syndrome, Addison’s disease, and rheu­
matoid arthritis) and those who reported cancer (n = 2,614 in the
2005–2006 survey, since in the follow-up study participants with this
condition were not included in the study). Although the main analyses in
2010 were restricted to participants who reported no infection at the
time of the interview (n = 526), a sensitivity analysis was performed on
the full population (n = 736) repeating the main analyses. Finally, we
also investigated possible differences by gender by including the inter­
action term with the contaminants in the models.
We expressed results as percentage difference (with a 95% confi­
dence interval – CI) in the outcomes associated with an interquartile
range (IQR) increment in PFAS concentrations. The percentage differ­
ence was calculated as the complement of the exponentiated regression
coefficient [(exp(β × IQR) – 1) × 100] for cell counts and just the
regression coefficient (β × IQR) for cell percentages. We used the
2005–2006 IQRs of ln(PFASs) for both populations to make results
comparable. For easier clinical interpretation, results for count data
were also expressed in terms of absolute differences, taking the median
values as the reference [median(cells) * (exp(β × IQR) – 1)]. Differences
in percentages of cells were evaluated to determine whether there were
differential effects of PFASs on this relative scale as opposed to the ef­
fects on absolute counts (e.g., an increase in total lymphocyte count
might not occur evenly across the different subsets).
We also re-ran the main analysis with deciles (2005–2006 popula­
tion) and tertiles (2010 population) of PFASs to explore the shape of the
exposure–response curve. To be consistent with the regression on the ln
(PFASs) scale, tests for trend were conducted across percentiles
including the ln-transformed geometric mean of each decile or tertile
added to the models as a numerical regressor. Results of categorical
analyses are presented graphically as the adjusted value of each
outcome, by exposure group, with 95% CI.
To be conservative, we used heteroscedasticity-consistent standard
errors in all the analyses to account for potential deviations from the
linear model assumptions (Long and Ervin, 2000). We conducted sta­
tistical analyses using R.3.6.2 (R Core Team, 2020).
3. Results
3.1. Study population, immune outcomes, and PFAS concentrations
The baseline population (2005–2006) was comprised of similar
numbers of women (53%) and men (47%), 29% of whom were smokers,
and 72% were overweight or obese. Significant differences between the
2005–2006 and 2010 surveys were found in some characteristics, with
higher educational levels, fewer smokers and alcohol drinkers, and
slightly older participants in 2010 (n = 526, mean age: 46 years)
compared to 2005–2006 (n = 42,782, mean age: 44 years) (Table 1).
PFHxS, PFOA, PFOS, and PFNA were detectable (above the LOD) in
97.4, 99.9, 99.5, and 97.6% of people in 2005–2006. Corresponding
data for 2010 were: 99.4, 99.8, 99.4, and 99.2% (data not shown in
table). The median serum PFASs in the follow-up compared to baseline
was higher for PFOA (due to the study population being located in the
six most contaminated water districts in 2010), lower for PFOS, similar
for PFHxS, and the same for PFNA (Table 1). In 2005–2006, the IQR
contrast for PFOA was bigger (56.0 ng/ml), PFOS was intermediate
(15.1 ng/ml), and both PFHxS and PFNA were smaller (2.8 and 0.7 ng/
ml, respectively). IQR contrasts for 2010 were: 69.8, 9.0, 2.0, and 0.7
ng/mL for PFOA, PFOS, PFHxS, and PFNA, respectively (data not shown
in table). Furthermore, positive correlations (p < 0.001 in all cases) were
found between all the PFASs analyzed (r range: 0.32–0.57 in 2005–2006
and 0.44–0.77 in 2010). We found weak inverse correlations between
PFASs and eGFR (r range from − 0.13 to − 0.03, with p < 0.05 except for
M.-J. Lopez-Espinosa et al. Environment International 156 (2021) 106599

Table 1 Table 2
Study population and PFAS concentrations, Mid-Ohio Valley, USA (2005–2010). Counts of WBCs and lymphocyte subsets, and association between PFASs and
Variable 2005–2006 2010
relative (percent) differences in these counts amongst the population in
2005–2006 (n = 42,782) and in 2010 (n = 526), Mid-Ohio Valley, USA.
(n ¼ 42,782) (n ¼ 526) pa
Sex 0.292 Cell PFHxS PFOA PFOS PFNA
Female 22,542 (52.7) 265 (50.4) count
Male 20,240 (47.3) 261 (49.6) Outcome Median % diff. % diff. % diff. % diff.
Race/ethnicity 0.071 (IQR) (95% (95% (95% (95%
Non-Hispanic white 41,735 (97.6) 519 (98.7) CI) CI) CI) CI)
Others 1,047 (2.45) 7 (1.33) Total WBCs 7,100 0.55 − 0.27 − 0.55 − 0.21
Age <0.001 2005–2006 (5,900, (0.25, (-0.62, (-0.84, (-0.47,
≤30 years 7,745 (18.1) 62 (11.8) 8,600) 0.86) 0.08) − 0.26) 0.05)
31–40 years 10,444 (24.4) 122 (23.2) Total WBCs 2010 6,300 0.90 0.84 0.55 − 0.08
41–50 years 11,804 (27.6) 132 (25.1) (5,300, (-1.47, (-2.20, (-1.35, (-1.95,
51–60 years 9,808 (22.9) 138 (26.2) 7,600) 3.33) 3.97) 2.49) 1.82)
>60 years 2,981 (6.97) 72 (13.7) Neutrophils 4,400 − 0.04 − 0.92 − 1.56 − 0.77
Education <0.001 2005–2006 (3,500, (-0.44, (-1.37, (-1.93, (-1.11,
<12 years 3,902 (9.12) 16 (3.04) 5,500) 0.35) − 0.47) − 1.19) − 0.44)
HSD or GED 17,461 (40.8) 158 (30.0) Neutrophils 2010 4,000 − 1.99 − 1.44 − 0.86 − 1.14
Some college 14,881 (34.8) 242 (46.0) (3,200, (-4.97, (-5.12, (-3.38, (-3.68,
≥Bachelor’s degree 6,538 (15.3) 110 (20.9) 5,000) 1.08) 2.39) 1.72) 1.46)
BMI 0.436 Monocytes 400 1.16 0.61 0.38 0.15
Underweight 486 (1.14) 6 (1.14) 2005–2006 (300, (0.71, (0.08, (-0.06, (-0.24,
Normal weight 11,607 (27.1) 139 (26.4) 500) 1.63) 1.15) 0.82) 0.54)
Overweight 14,826 (34.7) 206 (39.2) Monocytes 2010 300 1.85 4.29 1.41 0.55
Obese I 9,198 (21.5) 96 (18.3) (300, (-2.71, (-1.05, (-2.38, (-2.81,
Obese II and III 6,665 (15.6) 79 (15.0) 400) 6.63) 9.92) 5.35) 4.02)
Smoking <0.001 Eosinophils 100 − 0.96 − 0.06 − 0.29 − 0.76
Never smoked 20,017 (46.8) 331 (62.9) 2005–2006 (100, (-1.76, (-0.98, (-1.08, (-1.44,
Ex-smoker 10,509 (24.6) 111 (21.1) 200) − 0.15) 0.87) 0.50) − 0.07)
Current smoker 12,256 (28.6) 84 (16.0) Eosinophils 2010 100 1.15 4.77 4.22 6.37
Alcohol intake <0.001 (100, (-6.64, (-4.11, (-2.33, (0.68,
Lifetime non-drinker 7,898 (18.5) 303 (57.6) 200) 9.59) 14.46) 11.21) 12.38)
Ex-drinker 12,207 (28.5) 52 (9.89) Lymphocytes 2,000 2.00 1.12 1.95 1.47
Drinker 22,677 (53.0) 171 (32.5) 2005–2006 (1,700, (1.63, (0.70, (1.57, (1.15,
PFAS (ng/mL) 2,500) 2.37) 1.55) 2.33) 1.79)
PFHxS 2.90(1.80, 4.60) 2.10(1.20, 3.20) <0.001 Lymphocytes 2010 1,800 7.33 5.50 3.39 1.29
PFOA 26.9(13.2, 69.2) 35.7(15.0, 93.7) <0.001 (1,500, (4.12, (1.52, (0.70, (-1.04,
PFOS 19.7(13.3, 28.4) 9.60(6.10, 14.9) <0.001 2,200) 10.64) 9.63) 6.15) 3.68)
PFNA 1.40(1.10, 1.80) 1.40(1.10, 1.80) <0.001 CD3þ T cells 1,356 7.43 5.89 3.12 1.30
(1,085, (3.80, (1.46, (0.20, (-1.25,
Numbers (%) or median (interquartile range) are presented. BMI, Body mass
1,645) 11.19) 10.51) 6.13) 3.92)
index (only available for the 2005–2006 survey); GED, General education
CD3þCD4þT-helper 904 8.28 7.42 3.94 2.04
development; HSD, High school diploma. PFAS, Perfluoroalkyl substance. cells (714, (4.28, (2.78, (0.60, (-0.84,
PFHxS, Perfluorohexane sulfonate; PFNA, Perfluorononanoic acid; PFOA, Per­ 1105) 12.44) 12.26) 7.40) 5.01)
fluorooctanoic acid; PFOS, Perfluorooctane sulfonate. CD3þCD8þT- 401 5.51 3.99 1.62 − 0.88
a
Likelihood ratio test p-value from unadjusted ordinal mixed models (for cytotoxic cells (293, (0.52, (-2.40, (-2.31, (-3.97,
sociodemographic covariates), or linear mixed models adjusted for gender, age, 534) 10.75) 10.80) 5.71) 2.32)
smoking, month of sampling, alcohol intake, and educational level (for ln CD3þCD4þCD8þDP 16 (10, 6.03 12.21 − 0.23 1.06
(PFAS)). T cells 25) (-1.21, (2.47, (-5.68, (-4.01,
13.79) 22.89) 5.55) 6.40)
CD3¡ 184 8.62 4.94 5.40 0.31
PFOS and PFNA in 2010) (Figure S1). CD16þCD56þNK (129, (3.45, (-0.75, (1.35, (-3.34,
Most median cell counts were slightly lower in the later survey cells 260) 14.04) 10.96) 9.60) 4.10)
CD3¡ CD19þB cells 238 7.26 3.32 3.89 2.26
sample, though this was not the case for eosinophils (Table 2). Median
(170, (2.08, (-2.82, (-0.64, (-1.79,
percentages of the different immune cells were mostly very similar be­ 318) 12.70) 9.86) 8.63) 6.47)
tween the two surveys (Table 3). CD4þ/CD8þratio 2.35 2.63 3.29 2.28 2.94
(1.71, (-2.19, (-2.53, (-1.78, (-0.31,
3.02) 7.68) 9.46) 6.52) 6.30)
3.2. Serum PFASs and WBC count
Immune cell count > 0 in 100% of samples except for monocytes (99.9% and
For total WBC count in the 2005–2006 survey, positive (PFHxS) and 99.6% in 2005–2006 and 2010, respectively) and eosinophils (91.9% and
inverse (rest of the contaminants) associations were found, reaching 93.2%). PFASs and counts of immune cells were log transformed (ln) for ana­
statistical significance for PFHxS and PFOS (difference: 0.55% and lyses. Likelihood ratio test p-value < 0.001 for the comparison between the two
time points for all outcomes except eosinophils (linear mixed models). Cell
− 0.55% per PFHxS and PFOS IQR increment, corresponding to 39 and
counts are expressed as cells/μL. Results are expressed as the difference in the
− 39 cells/μl when referring to the median population count). WBC as­
outcomes associated with IQR increments in PFAS levels in 2005–2006. All
sociations showed wide confidence intervals in the smaller follow-up models were adjusted for gender, age, smoking, month of sampling, alcohol
population (Tables 2 and 3 and S1 and Fig. 1 and S2). intake, and educational level.
% diff., % difference; CI, Confidence interval; DP, Double positive; IQR, Inter­
3.3. Serum PFASs and myeloid lineage quartile range; NK, Natural killer; PFASs, Perfluoroalkyl substances; PFHxS,
Perfluorohexane sulfonate; PFNA, Perfluorononanoic acid; PFOA, Per­
There was some evidence of an association between PFOA, PFOS, fluorooctanoic acid; PFOS, Perfluorooctane sulfonate; WBCs, White blood cells.
PFNA, and neutrophil count (difference range: − 1.56% to − 0.77% per
PFAS IQR increment [–69 to –34 cells/μl]) and between the four PFASs
and neutrophil percentage (− 0.65 to − 0.36 per PFAS IQR increment) in

4
M.-J. Lopez-Espinosa et al. Environment International 156 (2021) 106599

Table 3 2010 survey. Finally, eosinophils were associated with PFHxS in the first
Percentages of WBCs and lymphocyte subsets, and association between PFASs survey (difference: − 0.96% and − 0.04 for count and percentage, per
and percentages of these cells amongst the population in 2005–2006 (n = PFHxS IQR increment, respectively) and PFNA in both surveys, but the
42,782) and in 2010 (n = 526), Mid-Ohio Valley, USA. direction of the latter association changed between the two time points
Cell PFHxS PFOA PFOS PFNA (difference: − 0.76% and 6.37% for count and − 0.03 and 0.13 for per­
percentage centage, per PFNA IQR increment) (Tables 2–3, and Fig. 1 and S2).
Outcome Median diff. diff. diff. diff. Given the smaller contribution of monocytes and eosinophils to the WBC
(IQR) (95% (95% (95% (95% count, such changes were almost negligible (range: 1–6 cells/μl) when
CI) CI) CI) CI) expressed in absolute terms (Table S1).
Neutrophils 62 (57, 67)a − 0.36 − 0.41 − 0.65 − 0.37
2005–2006 (-0.46, (-0.52, (-0.74, (-0.45,
− 0.26) − 0.30) − 0.55) − 0.29) 3.4. Serum PFASs and lymphoid lineage
Neutrophils 2010 63 (57, 68)a − 1.74 − 1.41 − 0.85 − 0.58
(-2.54, (-2.38, (-1.63, (-1.31, For total lymphocytes, positive associations were clearly evident for
− 0.94) − 0.44) − 0.07) 0.15)
PFHxS, PFOA, and PFOS. These associations were stronger for absolute
Monocytes 6.0 (5.0, 0.02 0.04 0.04 0.00
2005–2006 7.0)a (0.00, (0.02, (0.02, (-0.02, numbers compared to percentages and with a larger effect size in the
0.05) 0.07) 0.06) 0.02) 2010 subset compared to the 2005–2006 population. Specifically, the
Monocytes 2010 5.0 (4.0, 0.02 0.16 0.03 0.04 relative (percent) differences in the counts of lymphocytes for an IQR
7.0)a (-0.2, (-0.08, (-0.15, (-0.11, contrast in PFHxS was 2.00% in the large population and 7.33% in the
0.24) 0.39) 0.20) 0.20)
Eosinophils 2.0 (1.0, − 0.04 0.00 − 0.01 − 0.03
follow-up, 1.12% and 5.50% for PFOA, and 1.95% and 3.39% for PFOS
2005–2006 3.0)a (-0.06, (-0.02, (-0.03, (-0.04, (Table 2). Concerning absolute counts, these estimates represented
− 0.02) 0.02) 0.01) − 0.01) changes ranging from 22 to 132 cells/μl (Table S1). Respective data for
Eosinophils 2010 2.0 (1.0, − 0.04 0.05 0.06 0.13 the difference increase in lymphocyte percentage were 0.37 and 1.77
3.0)a (-0.23, (-0.21, (-0.11, (0.01,
(PFHxS), 0.36 and 1.24 (PFOA), and 0.61 and 0.77 (PFOS) (Table 3).
0.15) 0.32) 0.22) 0.26)
Lymphocytes 29 (24, 34)a 0.37 0.36 0.61 0.39 Fig. 1 and S2 show these graphically across PFAS deciles (population in
2005–2006 (0.28, (0.26, (0.53, (0.32, 2005–2006) and tertiles (population in 2010) with, in general, clear
0.46) 0.45) 0.69) 0.46) monotonic increases for these three contaminants and lymphocytes. A
Lymphocytes 2010 29 (24, 34)a 1.77 1.24 0.77 0.43 monotonic association was also found between PFNA and lymphocyte
(1.05, (0.36, (0.13, (-0.23,
2.50) 2.12) 1.42) 1.08)
cells (difference for count and percentage: 1.47% [29 cells/μl] and 0.39
CD3þT cells 75 (70, 79)b 0.02 0.27 − 0.20 0.00 per PFNA IQR increment) in the 2005–2006 population but not in that of
(-0.70, (-0.54, (-0.79, (-0.52, 2010 (Tables 2–3 and Fig. 1 and S2).
0.74) 1.07) 0.39) 0.52) The above-mentioned statistically significant changes in absolute
CD3þCD4þT-helper 50 (45, 55)b 0.45 0.88 0.24 0.40
lymphocyte count led to changes in the numbers of lymphocyte subsets.
cells (-0.38, (-0.05, (-0.45, (-0.15,
1.28) 1.82) 0.93) 0.95) Specifically, we found associations between most of the types of
CD3þCD8þT- 22 (17, 27)b − 0.44 − 0.36 − 0.51 − 0.42 numbers of lymphocyte subsets and PFHxS concentrations, with differ­
cytotoxic cells (-1.29, (-1.40, (-1.26, (-1.01, ences ranging from 5.51% in CD3+ CD8+ T-cytotoxic cells to 8.62% in
0.41) 0.68) 0.24) 0.17) CD3− CD16+ CD56+ NK cells per PFHxS IQR increment (Table 2).
CD3þCD4þCD8þDP 0.9 (0.6, − 0.03 0.15 − 0.05 0.02
T cells 1.3)b (-0.12, (-0.02, (-0.13, (-0.07,
Fig. 2 shows clear monotonic increases across PFHxS tertiles for all B,
0.07) 0.33) 0.03) 0.11) NK, and T cell numbers except the CD3+ CD4+ CD8+ DP subset. For
CD3¡ 10 (7.3, 0.16 0.04 0.27 − 0.04 other PFASs the associations were more limited. For PFOA, associations
CD16þCD56þNK 14)b (-0.37, (-0.62, (-0.12, (-0.44, with the numbers of the above-mentioned lymphocytes were also
cells 0.70) 0.71) 0.66) 0.37)
evident for CD3+ T cells (5.89%) and CD3+ CD4+ T-helper cells
CD3¡ CD19þB cells 13 (10, 17)b − 0.18 − 0.33 − 0.04 0.06
(-0.77, (-0.93, (-0.55, (-0.39, (7.42%), and more strongly for CD3+ CD4+ CD8+ DP T cells (12.21%).
0.42) 0.28) 0.46) 0.51) For PFOS, again there were associations with both CD3+ and CD3+
CD4+ T cells, and CD3− CD16+ CD56+ NK cells (3.12%, 3.94%, and
Immune cell count > 0 in 100% of samples except for monocytes (99.9% and
99.6% in 2005–2006 and 2010, respectively) and eosinophils (91.9% and
5.40%, respectively) (Table 2). The magnitudes of the associations
93.2%). PFASs were log transformed (ln) for analyses. Likelihood ratio test p- increased more clearly monotonically across tertiles for both CD3+ T
value < 0.001 for the comparison between both time points for all outcomes cells and CD3+ CD4+ T cells in relation to PFOA, and CD3− CD16+
except lymphocytes and eosinophils (linear mixed models). Results are CD56+ NK cells in relation to PFOS (Fig. 2). PFNA was not associated
expressed as the difference in the outcomes associated with IQR increments in with any subtype of lymphocyte (Table 2 and Fig. 2).
PFAS levels in 2005–2006. All models were adjusted for gender, age, smoking, Regarding percentage data, associations were mostly null in all cases,
month of sampling, alcohol intake, and educational level. meaning that changes in actual cell numbers strongly resembled those
CI, Confidence interval; diff., difference; DP, Double positive; IQR, Interquartile seen for total lymphocytes, but now distributed accordingly across the
range; NK, Natural killer; PFASs, Perfluoroalkyl substances; PFHxS, Per­ different phenotypes (Table 3 and Figure S3).
fluorohexane sulfonate; PFNA, Perfluorononanoic acid; PFOA, Per­
fluorooctanoic acid; PFOS, Perfluorooctane sulfonate.
a 3.5. Sensitivity analyses
Percentage of total WBCs.
b
Percentage of total lymphocytes.
Various models including other variables which might have
confounded the results are set out in detail in the Supplementary ma­
the first survey. The effect was notably larger for the percentage of
terial (Figures S4–S9). Firstly, modest rather than major changes in the
neutrophils in relation to PFHxS, PFOA, and PFOS in the follow-up
associations resulted after adding BMI, eGFR, or anti-inflammatory
population (difference: − 1.74, − 1.41, and − 0.85, per PFAS IQR incre­
drugs, or excluding self-reported cancer or autoimmune diseases. Sec­
ment, respectively). We also found some evidence of a positive associ­
ondly, in models containing all four PFASs, the regression coefficients
ation between PFHxS, PFOA, and monocyte count (difference: 1.16%
fell or were even close to null in some cases and their CIs were wider.
and 0.61% per PFAS IQR increment, respectively) and between PFHxS,
Specifically, for the myeloid lineage bigger changes were shown in the
PFOA, PFOS, and monocyte percentage (difference range: 0.02 to 0.04
2005–2006 survey compared to that of 2010. Regarding lymphocytes,
per PFAS IQR increment, respectively) in the 2005–2006 but not in the
the association was somewhat smaller for PFHxS versus count and null

5
M.-J. Lopez-Espinosa et al.
Fig. 1. Adjusted WBC counts (mean and 95% confidence interval) by PFAS deciles (2005–2006, n ¼ 42,782) or tertiles (2010, n ¼ 526), Mid-Ohio Valley,
USA. Statistically significant trends at p < 0.05 and p < 0.01 are denoted by one and two asterisks, respectively. All models were adjusted for gender, age, smoking,
month of sampling, alcohol intake, and educational level. PFAS, Perfluoroalkyl substance; PFHxS, Perfluorohexane sulfonate; PFNA, Perfluorononanoic acid; PFOA,
Perfluorooctanoic acid; PFOS, Perfluorooctane sulfonate, WBCs, White blood cells.
for percentage in the 2005–2006 survey. For PFOA, the associations
were smaller, reaching the null in 2005-2005 for count and losing sta­
tistical significance for count and percentage in 2010. For PFOS and both
count and percentage the associations were smaller in 2005–2006 and
disappeared in 2010. For numbers of lymphocyte subtypes including the
four PFASs simultaneously, the associations were, in general, a little
smaller for PFHxS and PFOA but completely disappeared for PFOS.
Thirdly, multivariate analysis amongst the full follow-up population, i.
e., population reporting (or not) infections in 2010 (n = 736) also sug­
gested associations between PFHxS, PFOA, and PFOS in relation to im­
mune cell counts, but with generally lower % differences and some with
CIs including the null. Finally, although some p-values of the interaction
with sex were < 0.05, especially for the WBC types, no clear evidence of
a differential effect between males and females was found on consid­
ering the two surveys.
4. Discussion
We determined peripheral WBC counts in two surveys (2005–2006
and 2010) and lymphocyte subtypes in 2010 to characterize potential
health effects related to the immune system associated with PFAS con­
centrations. Our community-based sample had a wide range of PFOA
serum concentrations (IQR: 13.2–69.2 ng/mL; median: 26.9 ng/mL in
2005–2006), and background levels of PFHxS, PFOS, and PFNA (me­
dians: 2.9, 19.7, and 1.4 ng/mL in 2005–2006, respectively). The trends
with PFASs for WBCs were slight, with small percentage differences
across the IQRs. For most PFASs there were negative slopes for neutro­
phils, with larger positive slopes for lymphocytes. Our results suggest a
consistent positive association between serum PFHxS and absolute
lymphocyte count and weaker evidence of an association for percentage
of total lymphocytes. For PFHxS and absolute counts of lymphocyte
subtypes, significant and positive associations were found for T, B, and
NK cell counts, which likely reflect the changes in the absolute
lymphocyte count. A less consistent association was also shown for both
PFOA and PFOS and total and subtypes of lymphocytes.
6
Environment International 156 (2021) 106599
No clear association was found between total WBC count and PFAS
levels, statistical significance only being reached for PFHxS and PFOS in
the 2005–2006 survey. The effects were minimal, and the direction of
the association was not consistent with PFHxS, showing a positive as­
sociation and the others showing negative trends. Two studies
measuring serum PFOA and PFOS in workers at 3M (n = 53) and DuPont
(n = 518) facilities located in Italy, Alabama (USA), and Belgium did not
report any significant associations between these contaminants and total
WBC count (range of PFOA and PFOS means were much higher than in
the present study: 0.07–19.7 and 0.13–1.40 μg/mL, respectively) (Costa
et al., 2009; Olsen et al., 2003). Diverging results have been reported in
non-occupational studies. Specifically, in a study on a population (n =
371) with longstanding environmental PFOA exposure (median: 354
ng/mL) in the same area as the present study, marginally significant
positive associations with WBC count were found (Emmett et al., 2006).
A cross-sectional study in pregnant women (n = 189) in Greenland
(Knudsen et al., 2018) reported a significant negative association be­
tween the sum of 17 PFASs (ΣPFAS median: 18.5 ng/mL) and WBC
count. However, levels of 4 PFASs in maternal and child samples at 18
months and 5 years of age (ΣPFAS geometric mean range: 0.47–0.73 ng/
mL) were not associated with WBC count at 5 years of age in a cohort (n
= 56) of Faroese children (Oulhote et al., 2017).
Examination of cell differentials resulted in mixed associations for
the myeloid lineage. Neutrophils showed slightly decreased trends
associated with PFASs in the first survey and less clearly in the second
survey, where the association was only with percentages of this cell type.
In line with our results, negative associations were reported between the
sum of 17 PFASs and numbers of neutrophils in the above-mentioned
pregnant women from Greenland (Knudsen et al., 2018). Conversely,
other studies did not find such an association (Dong et al., 2013; Emmett
et al., 2006; Oulhote et al., 2017). Regarding other myeloid cells, we
found positive associations between PFHxS, PFOA, and PFOS and
monocytes in the first survey, while no consistent pattern of an associ­
ation was found for monocytes in the second survey or for eosinophils. In
the above-mentioned study on the Mid-Ohio Valley, significant positive
M.-J. Lopez-Espinosa et al.
Fig. 2. Adjusted lymphocyte subtype counts (mean and 95% confidence
interval) by PFAS tertiles (n ¼ 526), Mid-Ohio Valley, USA (2010). Sta­
tistically significant trends at p < 0.05 and p < 0.01 are denoted by one and two
asterisks, respectively. All models were adjusted for gender, age, smoking,
month of sampling, alcohol intake, and educational level. DP, Double positive;
NK, Natural killer; PFAS, Perfluoroalkyl substance; PFHxS, Perfluorohexane
sulfonate; PFNA, Perfluorononanoic acid; PFOA, Perfluorooctanoic acid; PFOS,
Perfluorooctane sulfonate.
associations with monocyte count but not percentage were found
(Emmett et al., 2006). Interestingly, none of the above-mentioned im­
mune cell types measured in children aged 5 years were associated with
prenatal, 18-month or 5-year concentrations of PFASs in Faroese chil­
dren (Oulhote et al., 2017). Cross-sectional studies reported conflicting
results for eosinophil counts, with one study finding a positive associa­
tion with PFOA (median in cases [n = 231] and controls [n = 225]: 1.2
and 0.5 ng/mL, respectively) and PFOS (33.9 and 28.9 ng/mL) among
asthmatic compared to nonasthmatic Taiwanese teenagers (Dong et al.,
2013). However, another study reported null results for 6 PFAS conge­
ners (median ΣPFASs for cases [n = 118] and controls [n = 169]: 7.3 and
5.7 ng/mL) among New Yorkers aged 16–17 years exposed to high levels
of PFASs, among other contaminants, in the World Trade Center disaster
who were matched to a control group (Gaylord et al., 2019).
Results were clearer for total lymphocytes, showing statistically
significant positive and generally monotonic associations with all PFAS
levels except PFNA in the two surveys. Specifically, PFHxS presented the
strongest effect for a small absolute contrast in levels (i.e., for an IQR
contrast of only 2.8 ng/mL it showed the biggest increase with both
count and percentage of lymphocytes). For PFOA and PFOS, there was
also an association, but less clear. With regard to the previous epide­
miological studies measuring lymphocytes, none have reported PFAS
exposure as being associated with alterations in this particular immune
cell type during childhood (Abraham et al., 2020; Oulhote et al., 2017)
or adulthood (Emmett et al., 2006; Knudsen et al., 2018).
To better characterize the lymphocytosis and help determine
whether there was any evidence of altered immunoregulation, immu­
nophenotyping of peripheral blood was conducted in the 2010 survey to
enumerate the major lymphocyte populations. For lymphocyte sub­
types, PFHxS concentrations were positively and monotonically associ­
ated with absolute counts of T, B, and NK cells. In addition, higher PFOA
serum levels were positively associated with some T cell subsets and
PFOS with some T cell subsets and NK cells, with monotonicity shown in
7
Environment International 156 (2021) 106599
some cases. Since counts of lymphocyte subtypes are not independent of
the absolute lymphocyte count and there were only significant changes
in counts but not in percentages of the lymphocyte subsets, the observed
changes likely reflect changes in absolute lymphocyte counts rather than
any specific effect on lymphocyte subsets themselves. Something similar
has been reported in smoker’s leukocytosis, a well-documented effect of
active smoking (Pedersen et al., 2019; Sopori, 2002) where the effect of
smoking raises WBCs without greatly changing subset percentages of the
different cell types (Park et al., 1992). With regard to PFASs, only two
epidemiological studies, both focused on child populations, have been
conducted on peripheral blood lymphocytes and PFASs, and no signifi­
cant associations were reported (Abraham et al., 2020; Oulhote et al.,
2017). On a different note, animal experiments have reported altered
lymphocyte populations following exposure to PFASs, although effects
seem species- or even strain-dependent (DeWitt et al., 2009). Increased
CD3+ CD4+ T-helper cells and decreased CD3+ CD8+ T-cytotoxic cells
in spleens and increased CD3+ CD8+ T–cytotoxic cells in thymus were
reported in male ICR mice exposed to PFOA for 21 days (Son et al.,
2009). Reduced B and T cells were also reported, particularly CD4+
CD8+ thymocytes in male C57BL/6 mice treated with PFOA for 7 or 10
days (Yang et al., 2001). In contrast, our finding of a positive association
between PFAS exposure and CD3− CD16+ CD56+ NK cells might be
more consistent with the increase in NK cell activity reported in B6C3F1
male mice exposed to PFOS (Peden-Adams et al., 2008).
In the present study, despite the evidence of an association with
increased total lymphocyte cells, the absolute magnitudes of difference
in cell numbers were small (difference range: 1.12–7.33% per PFAS IQR
increment, which when compared to the population median represented
at most 132 [95% CI: 74, 191] cells/μL for PFHxS in 2010) and the
percentages of total WBCs that these cells represented changed very
little (difference range: 0.36–1.77 per PFAS IQR increment). Therefore,
lymphocyte changes in the present study cannot be considered clinically
meaningful. The same is true for the decreases observed in absolute
neutrophil count: though statistically significant for some PFASs, espe­
cially for the 2005–2006 data, these were quite small and would not be
clinically important in themselves. Further studies would be needed to
determine whether these effects could be meaningful in specific clinical
contexts, such as a serious bacterial infection.
Overall, these findings, along with previous findings of suppressed
antibody response (Abraham et al., 2020; Grandjean et al., 2017, 2012;
Kielsen et al., 2016; Looker et al., 2014), suggest that the effect of PFASs
on immune function, if there is one, is not mediated solely through
altered lymphocyte numbers but by changes in cell signaling or function
(as measured by antibody response).
Previous studies in the population considered here have suggested
that impaired kidney function may lead to decreased net PFOA excretion
via the kidneys and, accordingly, increased serum levels of PFOA
(Watkins et al., 2013). Thus, confounding by kidney function may
plausibly occur due to the fact that inflammation or ill health, which can
be associated with renal impairment, may also be reflected in the
elevated lymphocyte count. Since eGFR reflects the filtering capacity of
the kidney (Dhingra et al., 2017; Watkins et al., 2013) and was available
for the study population, we repeated the regression analysis including
eGFR in the models, with no evidence of eGFR confounding the asso­
ciation between PFASs and immune cell outcomes. Estimates in the main
analysis did not differ markedly when adding other possible con­
founders such as 2005–2006 BMI or self-reported anti-inflammatory
drug therapy (only available for the follow-up study). The main differ­
ences were found for the inclusion of the four PFASs at the same time
since CIs were wider, and the estimates fell when including other PFASs
or were even close to null in some cases. The exclusion of people who
reported autoimmune disease (both surveys) or cancer (first survey)
hardly modified the estimated associations. Finally, gender-specific
differences in the development of reproductive organs and physiology
may result in gender-specific responses to a given toxicant (Bach et al.,
2015). For PFASs, two studies have indicated a higher vulnerability
M.-J. Lopez-Espinosa et al.
among asthmatic boys compared to girls (Qin et al., 2017; Zhu et al.,
2016). Mixed results were reported in another study with girls or boys at
higher risk depending on the outcome (asthma, atopic dermatitis, and
rhinitis) and the class of PFAS studied (Kvalem et al., 2020). However,
although we found some significant gender interactions indicating that
higher PFAS exposure was associated with higher counts of types of
WBCs in females, and lower counts in males, such a pattern was not
consistent.
Our study has several limitations. First, information on residential
histories, infections, and other potential confounders was self-reported,
and participants were only asked about the presence or absence (not
severity) of specific infections. Nevertheless, restricting our analysis to
those who reported no infections in the 2010 survey should have
reduced potential bias due to infection reports. Second, information on
infections was only available for the follow-up population and BMI only
for the first survey. Nevertheless, the results for participants who did not
report infection in 2010 (n = 526) are somewhat stronger than those for
the full population in 2010 (n = 736) or for the 2005–2006 cohort. In
addition, sensitivity analyses including 2005–2006 BMI did not sub­
stantially change the results in either of the two surveys, thus ruling out
any role of BMI as a confounder. Furthermore, in this population PFOA
exposure was largely driven by residential water district and consump­
tion, and less correlated with individual self-reported characteristics and
behaviors (Steenland et al., 2009). Third, given the high-exposure na­
ture of our population for PFOA, the results for this contaminant may not
be generalizable to other populations. Fourth, the immune system has
many components but we studied only the absolute counts and per­
centages of the main types of immune cells, with no assessment of the
function of these cells. However, it is pertinent to highlight that, in a
previous study of part of the follow-up population of the present study,
increased PFOA concentrations were associated with a lower antibody
response to A/H3N2 influenza vaccine (Looker et al., 2014). Fifth, due to
the high number of statistical analyses, there is the possibility of po­
tential false positive statistically significant associations (i.e., type I er­
rors). The estimates for the coefficients and their confidence intervals
should be taken as a global picture of the pattern of the relations be­
tween the variables involved in the study. Finally, humans are poten­
tially exposed to a large variety of other types of contaminants, and
further study of the relation between exposures to multiple chemicals
with immunotoxic capability is warranted.
The major strengths of this study are the large sample size, assess­
ments of PFAS levels and some immune cells at two time points with an
interval of approximately five years, and the inclusion of individual
potential confounders. In addition, it is believed that these participants
are representative, at least for the first survey, given the high partici­
pation rates in the C8 Health Project (~85% in the range of age of the
present study) (Frisbee et al., 2009) of those adults who drank
contaminated water in the Mid-Ohio Valley, and this diminishes concern
about potential selection biases. Thirdly, despite an increasing body of
literature on the possible immunotoxic effects of PFASs in humans and
specific animal models demonstrating specific lymphocyte changes, to
our knowledge studies involving human immunophenotypes are scarce
and there is therefore a gap in the literature on this topic.
In summary, the strongest association was observed between abso­
lute lymphocyte count and PFHxS. If causal, our results suggest that
PFHxS may be more potent compared to the other PFASs, since a
contrast of only 2.8 ng/ml was associated with as much as a 7% increase
in absolute lymphocyte count in the 2010 survey. Weaker evidence of an
association was also shown for PFHxS and percentage of lymphocytes.
For PFOA and PFOS, the association was less clear. For all PFASs, the
magnitudes of the differences in absolute lymphocyte count were small.
A similar pattern was found for lymphocyte cell subtypes with, again,
stronger evidence for PFHxS. Nevertheless, the changes observed in
absolute counts of lymphocyte subsets reflected the change in total
lymphocyte count rather than any specific effect on the subsets them­
selves, since no association was reported for percentage of the subtypes.
8
Environment International 156 (2021) 106599
CRediT authorship contribution statement
Maria-Jose Lopez-Espinosa: Conceptualization, Data curation,
Formal analysis, Investigation, Methodology, Software, Validation,
Visualization, Writing - original draft, Writing - review & editing.
Christian Carrizosa: Data curation, Formal analysis, Investigation,
Methodology, Software, Visualization, Writing - original draft, Writing -
review & editing. Michael I. Luster: Conceptualization, Investigation,
Methodology, Resources, Validation, Validation, Writing - review &
editing. Joseph B. Margolick: Investigation, Methodology, Resources,
Visualization, Writing - original draft, Writing - review & editing. Olga
Costa: Data curation, Formal analysis, Investigation, Methodology,
Software, Writing - review & editing. Giovanni S. Leonardi: Concep­
tualization, Funding acquisition, Investigation, Methodology, Valida­
tion, Writing - review & editing. Tony Fletcher: Conceptualization,
Data curation, Formal analysis, Funding acquisition, Investigation,
Methodology, Project administration, Resources, Supervision, Valida­
tion, Visualization, Writing - original draft, Writing - review & editing.
Declaration of Competing Interest
The authors declare that they have no known competing financial
interests or personal relationships that could have appeared to influence
the work reported in this paper.
Acknowledgments
The authors wish to thank the participants for their contributions to
this study. We are also grateful to the Centers for Disease Control and
Prevention (CDC) laboratory for analyzing the PFASs in serum samples.
This project was funded by the C8 Class Action Settlement Agree­
ment (Circuit Court of Wood County, WV) between DuPont and plain­
tiffs, which resulted from releases of perfluorooctanoate (PFOA, or C8)
into drinking water. It was one of the C8 Science Panel Studies under­
taken by the Court-approved C8 Science Panel, of which Dr Tony
Fletcher was a member, established under the same Settlement Agree­
ment. Dr Fletcher is a consortium partner of the European Union’s Ho­
rizon 2020 research and innovation program under grant agreement
733032 HBM4EU (www.HBM4EU.eu); part-funded by the National
Institute for Health Research (NIHR) Health Protection Research Unit in
Environmental Exposures and Health, a partnership between Public
Health England, the Health and Safety Executive, and the University of
Leicester; and part funded with a PFAS research grant from CORIS/
REGIONE VENETO (Italy). The views expressed are those of the author
(s) and not necessarily those of the NIHR, Public Health England, the
Health and Safety Executive or the Department of Health and Social
Care. MJ Lopez-Espinosa holds grants from the Spanish Carlos III Health
Institute (Miguel Servet-FSE: MSII16/00051 and FIS-FEDER: PI14/
00891 and PI17/00663), Alicia Koplowitz Foundation 2017, and Min­
istry of Education, Culture and Sports (Jose Castillejo Grant: CAS17/
00052).
Appendix A. Supplementary material
Supplementary data to this article can be found online at https://doi.
org/10.1016/j.envint.2021.106599.
References
Abraham, K., Mielke, H., Fromme, H., Völkel, W., Menzel, J., Peiser, M., Zepp, F.,
Willich, S.N., Weikert, C., 2020. Internal exposure to perfluoroalkyl substances
(PFASs) and biological markers in 101 healthy 1-year-old children: associations
between levels of perfluorooctanoic acid (PFOA) and vaccine response. Arch.
Toxicol. 94, 2131–2147. https://doi.org/10.1007/s00204-020-02715-4.
Agency for Toxic Substances and Disease Registry [ATSDR], 2018. Toxicological profile
for perfluoroalkyls (Draft for public comment). Atlanta, GA: US Department of
Health and Human Services, Public Health Service.
M.-J. Lopez-Espinosa et al.
Aziz, N., Margolick, J.B., Detels, R., Rinaldo, C.R., Phair, J., Jamieson, B.D., Butch, A.W.,
2013. Value of a quality assessment program in optimizing cryopreservation of
peripheral blood mononuclear cells in a multicenter study. Clin. Vaccine Immunol.
20, 590–595. https://doi.org/10.1128/CVI.00693-12.
Bach, C.C., Bech, B.H., Brix, N., Nohr, E.A., Bonde, J.P.E., Henriksen, T.B., 2015.
Perfluoroalkyl and polyfluoroalkyl substances and human fetal growth: A systematic
review. Crit. Rev. Toxicol. 45, 53–67. https://doi.org/10.3109/
10408444.2014.952400.
Ballesteros, V., Costa, O., Iñiguez, C., Fletcher, T., Ballester, F., Lopez-Espinosa, M.J.,
2017. Exposure to perfluoroalkyl substances and thyroid function in pregnant
women and children: A systematic review of epidemiologic studies. Environ. Int. 99,
15–28. https://doi.org/10.1016/j.envint.2016.10.015.
Braun, J.M., 2017. Early-life exposure to EDCs: Role in childhood obesity and
neurodevelopment. Nat. Rev. Endocrinol. 13, 161–173. https://doi.org/10.1038/
nrendo.2016.186.
Cardenas, A., Hauser, R., Gold, D.R., Kleinman, K.P., Hivert, M.F., Fleisch, A.F., Lin, P.I.
D., Calafat, A.M., Webster, T.F., Horton, E.S., Oken, E., 2018. Association of
Perfluoroalkyl and Polyfluoroalkyl Substances With Adiposity. JAMA Netw. open 1,
e181493. https://doi.org/10.1001/jamanetworkopen.2018.1493.
Castilho, J.L., Shepherd, B.E., Koethe, J., Turner, M., Bebawy, S., Logan, J., Rogers, W.B.,
Raffanti, S., Sterling, T.R., 2016. CD4+/CD8+ratio, age, and risk of serious
noncommunicable diseases in HIV-infected adults on antiretroviral therapy. AIDS
30, 899–908. https://doi.org/10.1097/QAD.0000000000001005.
Chen, Q., Huang, R., Hua, L., Guo, Y., Huang, L., Zhao, Y., Wang, X., Zhang, J., 2018.
Prenatal exposure to perfluoroalkyl and polyfluoroalkyl substances and childhood
atopic dermatitis: A prospective birth cohort study. Environ. Heal. 17, 8. https://doi.
org/10.1186/s12940-018-0352-7.
Costa, G., Sartori, S., Consonni, D., 2009. Thirty years of medical surveillance in
perfluooctanoic acid production workers. J. Occup. Environ. Med. 51, 364–372.
https://doi.org/10.1097/JOM.0b013e3181965d80.
Davis, A.S., Viera, A.J., Mead, M.D., 2014. Leukemia: An overview for primary care. Am.
Fam. Physician 89, 731–738.
DeWitt, J.C., Shnyra, A., Badr, M.Z., Loveless, S.E., Hoban, D., Frame, S.R., Cunard, R.,
Anderson, S.E., Meade, B.J., Peden-Adams, M.M., Luebke, R.W., Luster, M.I., 2009.
Immunotoxicity of perfluorooctanoic acid and perfluorooctane sulfonate and the role
of peroxisome proliferator-activated receptor alpha. Crit. Rev. Toxicol. https://doi.
org/10.1080/10408440802209804.
Dhingra, R., Winquist, A., Darrow, L.A., Klein, M., Steenland, K., 2017. A study of reverse
causation: Examining the associations of perfluorooctanoic acid serum levels with
two outcomes. Environ. Health Perspect. 125, 416–421. https://doi.org/10.1289/
EHP273.
Dong, G.H., Tung, K.Y., Tsai, C.H., Liu, M.M., Wang, D., Liu, W., Jin, Y.H., Hsieh, W.S.,
Lee, Y.L., Chen, P.C., 2013. Serum polyfluoroalkyl concentrations, asthma outcomes,
and immunological markers in a case-control study of Taiwanese children. Environ.
Health Perspect. 121, 507–513. https://doi.org/10.1289/ehp.1205351.
EFSA [European Food Safety Authority], 2018. EFSA CONTAM Panel (EFSA Panel on
Contaminants in the Food Chain), Knutsen HK, Alexander J, Barregard L, Bignami M,
Bruschweiler B, Ceccatelli S, Cottrill B, Dinovi M, Edler L, Grasl-Kraupp B, Hogstrand
C, Hoogenboom LR, Nebbia CS, Oswald IP, Petersen A. R. FSA J. 16, 5194. https://
doi.org/10.2903/j.efsa.2018.5194.
Emmett, E.A., Zhang, H., Shofer, F.S., Freeman, D., Rodway, N.V., Desai, C., Shaw, L.M.,
2006. Community exposure to perfluorooctanoate: Relationships between serum
levels and certain health parameters. J. Occup. Environ. Med. 48, 771–779. https://
doi.org/10.1097/01.jom.0000233380.13087.37.
Fitz-Simon, N., Fletcher, T., Luster, M.I., Steenland, K., Calafat, A.M., Kato, K.,
Armstrong, B., 2013. Reductions in serum lipids with a 4-year decline in serum
perfluorooctanoic acid and perfluorooctanesulfonic acid. Epidemiology 24,
569–576. https://doi.org/10.1097/EDE.0b013e31829443ee.
Flaherty, J.M., Connolly, P.D., Decker, E.R., Kennedy, S.M., Ellefson, M.E., Reagen, W.K.,
Szostek, B., 2005. Quantitative determination of perfluorooctanoic acid in serum and
plasma by liquid chromatography tandem mass spectrometry. J. Chromatogr. B
Anal. Technol. Biomed. Life Sci. 819, 329–338. https://doi.org/10.1016/j.
jchromb.2005.03.002.
Frisbee, S.J., Brooks, A.P., Maher, A., Flensborg, P., Arnold, S., Fletcher, T., Steenland, K.,
Shankar, A., Knox, S.S., Pollard, C., Halverson, J.A., Vieira, V.M., Jin, C., Leyden, K.
M., Ducatman, A.M., 2009. The C8 Health Project: Design, methods, and
participants. Environ. Health Perspect. 117, 1873–1882. https://doi.org/10.1289/
ehp.0800379.
Gaylord, A., Berger, K.I., Naidu, M., Attina, T.M., Gilbert, J., Koshy, T.T., Han, X.,
Marmor, M., Shao, Y., Giusti, R., Goldring, R.M., Kannan, K., Trasande, L., 2019.
Serum perfluoroalkyl substances and lung function in adolescents exposed to the
World Trade Center disaster. Environ. Res. 172, 266–272. https://doi.org/10.1016/
j.envres.2019.02.024.
George-Gay, B., Parker, K., 2003. Understanding the complete blood count with
differential. J. Perianesthesia Nurs. 18, 96–117. https://doi.org/10.1053/
jpan.2003.50013.
Giorgi, J.V., Cheng, H.L., Margolick, J.B., Bauer, K.D., Ferbas, J., Waxdal, M., Schmid, I.,
Hultin, L.E., Jackson, A.L., Park, L., Taylor, J.M.G., 1990. Quality control in the flow
cytometric measurement of T-lymphocyte subsets: The Multicenter AIDS Cohort
Study experience. Clin. Immunol. Immunopathol. 55, 173–186. https://doi.org/
10.1016/0090-1229(90)90096-9.
Grandjean, P., Andersen, E.W., Budtz-Jørgensen, E., Nielsen, F., Mølbak, K.R., Weihe, P.,
Heilmann, C., 2012. Serum vaccine antibody concentrations in children exposed to
perfluorinated compounds. JAMA 307, 391–397. https://doi.org/10.1001/
jama.2011.2034.
9
Environment International 156 (2021) 106599
Grandjean, P., Heilmann, C., Weihe, P., Nielsen, F., Mogensen, U.B., Budtz-Jørgensen, E.,
2017. Serum vaccine antibody concentrations in adolescents exposed to
perfluorinated compounds. Environ. Health Perspect. 125, 077018 https://doi.org/
10.1289/EHP275.
Haase, H., Fahlenkamp, A., Schettgen, T., Esser, A., Gube, M., Ziegler, P., Kraus, T.,
Rink, L., 2016. Immunotoxicity monitoring in a population exposedto
polychlorinated biphenyls. Int. J. Environ. Res. Public Health 13, 1–13. https://doi.
org/10.3390/ijerph13030295.
International Pollutants Elimination Network, 2019. PFAS pollution across the Middle
East and Asia [WWW Document]. accessed 6.8.20. https://www.documentcloud.
org/documents/5980916-Pfas-Pollution-Across-the-Middle-East-and-Asia.html.
Kato, K., Basden, B.J., Needham, L.L., Calafat, A.M., 2011. Improved selectivity for the
analysis of maternal serum and cord serum for polyfluoroalkyl chemicals.
J. Chromatogr. A 1218, 2133–2137. https://doi.org/10.1016/j.
chroma.2010.10.051.
Kielsen, K., Shamim, Z., Ryder, L.P., Nielsen, F., Grandjean, P., Budtz-Jørgensen, E.,
Heilmann, C., 2016. Antibody response to booster vaccination with tetanus and
diphtheria in adults exposed to perfluorinated alkylates. J. Immunotoxicol. 13,
270–273. https://doi.org/10.3109/1547691X.2015.1067259.
Knudsen, A.K.S., Long, M., Pedersen, H.S., Bonefeld-Jørgensen, E.C., 2018. Persistent
organic pollutants and haematological markers in Greenlandic pregnant women: the
ACCEPT sub-study. Int. J. Circumpolar Health 77, 1456303. https://doi.org/
10.1080/22423982.2018.1456303.
Kvalem, H.E., Nygaard, U.C., Lødrup Carlsen, K.C., Carlsen, K.H., Haug, L.S., Granum, B.,
2020. Perfluoroalkyl substances, airways infections, allergy and asthma related
health outcomes – implications of gender, exposure period and study design.
Environ. Int. 134, 105259 https://doi.org/10.1016/j.envint.2019.105259.
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. https://doi.org/10.1093/toxsci/kfm128.
Leonardi, G.S., Houthuijs, D., Steerenberg, P.A., Fletcher, T., Armstrong, B., Antova, T.,
Lochman, I., Lochmanova, A., Rudnai, P., Erdei, E., Musial, J., Jazwiec-Kanyion, B.,
Niciu, E.M., Durbaca, S., Fabianova, E., Koppova, K., Lebret, E., Brunekreef, B., Van
Loveren, H., 2000. Immune biomarkers in relation to exposure to particulate matter:
A cross-sectional survey in 17 cities of Central Europe. Inhal. Toxicol. 12, 1–14.
https://doi.org/10.1080/089583700750019486.
Levey, A.S., Coresh, J., Greene, T., Stevens, L.A., Zhang, Y., Hendriksen, S., Kusek, J.W.,
Van Lente, F., 2006. Using standardized serum creatinine values in the Modification
of Diet in Renal Disease study equation for estimating glomerular filtration rate.
Ann. Intern. Med. 145, 247–254. https://doi.org/10.7326/0003-4819-145-4-
200608150-00004.
Long, J.S., Ervin, L.H., 2000. Using heteroscedasticity consistent standard errors in the
linear regression model. Am. Stat. 54, 217–224. https://doi.org/10.1080/
00031305.2000.10474549.
Looker, C., Luster, M.I., Calafat, A.M., Johnson, V.J., Burleson, G.R., Burleson, F.G.,
Fletcher, T., 2014. Influenza vaccine response in adults exposed to
perfluorooctanoate and perfluorooctanesulfonate. Toxicol. Sci. 138, 76–88. https://
doi.org/10.1093/toxsci/kft269.
Martí, A., Marcos, A., Martínez, J.A., 2001. Obesity and immune function relationships.
Obes. Rev. 2, 131–140. https://doi.org/10.1046/j.1467-789x.2001.00025.x.
Matilla-Santander, N., Valvi, D., Lopez-Espinosa, M.J., Manzano-Salgado, C.B.,
Ballester, F., Ibarluzea, J., Santa-Marina, L., Schettgen, T., Guxens, M., Sunyer, J.,
Vrijheid, M., 2017. Exposure to perfluoroalkyl substances and metabolic outcomes in
pregnant women: Evidence from the Spanish INMA birth cohorts. Environ. Health
Perspect. 125, 117004 https://doi.org/10.1289/EHP1062.
Olsen, G.W., Burris, J.M., Burlew, M.M., Mandel, J.H., 2003. Epidemiologic assessment
of worker serum perfluorooctanesulfonate (PFOS) and perfluorooctanoate (PFOA)
concentrations and medical surveillance examinations. J. Occup. Environ. Med. 45,
260–270. https://doi.org/10.1097/01.jom.0000052958.59271.10.
Oulhote, Y., Shamim, Z., Kielsen, K., Weihe, P., Grandjean, P., Ryder, L.P., Heilmann, C.,
2017. Children’s white blood cell counts in relation to developmental exposures to
methylmercury and persistent organic pollutants. Reprod. Toxicol. 68, 207–214.
https://doi.org/10.1016/j.reprotox.2016.08.001.
Park, L.P., Margolick, J.B., Giorgi, J.V., Ferbas, J., Bauer, K., Kaslow, R., Muñoz, A.,
1992. Influence of HIV-1 infection and cigarette smoking on leukocyte profiles in
homosexual men. The Multicenter AIDS Cohort Study. J. Acquir. Immune Defic.
Syndr. 5, 1124–1130.
Peden-Adams, M.M., Keller, J.M., Eudaly, J.G., Berger, J., Gilkeson, G.S., Keil, D.E.,
2008. Suppression of humoral immunity in mice following exposure to
perfluorooctane sulfonate. Toxicol. Sci. 104, 144–154. https://doi.org/10.1093/
toxsci/kfn059.
Pedersen, K.M., Çolak, Y., Ellervik, C., Hasselbalch, H.C., Bojesen, S.E., Nordestgaard, B.
G., 2019. Smoking and increased white and red blood cells. Arterioscler. Thromb.
Vasc. Biol. 39, 965–977. https://doi.org/10.1161/ATVBAHA.118.312338.
Qin, X. Di, Qian, Z. (Min), Dharmage, S.C., Perret, J., Geiger, S.D., Rigdon, S.E., Howard,
S., Zeng, X.W., Hu, L.W., Yang, B.Y., Zhou, Y., Li, M., Xu, S.L., Bao, W.W., Zhang, Y.
Z., Yuan, P., Wang, J., Zhang, C., Tian, Y.P., Nian, M., Xiao, X., Chen, W., Lee, Y.L.,
Dong, G.H., 2017. Association of perfluoroalkyl substances exposure with impaired
lung function in children. Environ. Res. 155, 15–21. https://doi.org/10.1016/j.
envres.2017.01.025.
R Core Team, 2020. R: A language and Environment for Statistical Computing. R
Foundation for Statistical Computing. Austria, Vienna.
Schenker, E.L., Hultin, L.E., Bauer, K.D., Ferbas, J., Margolick, J.B., Giorgi, J.V., 1993.
Evaluation of a dual-color flow cytometry immunophenotyping panel in a
multicenter quality assurance program. Cytometry 14, 307–317. https://doi.org/
10.1002/cyto.990140311.
M.-J. Lopez-Espinosa et al.
Son, H.Y., Lee, S., Tak, E.N., Cho, H.S., Shin, H.I., Kim, S.H., Yang, J.H., 2009.
Perfluorooctanoic acid alters T lymphocyte phenotypes and cytokine expression in
mice. Environ. Toxicol. 24, 580–588. https://doi.org/10.1002/tox.20459.
Sopori, M., 2002. Effects of cigarette smoke on the immune system. Nat. Rev. Immunol.
2, 372–377. https://doi.org/10.1038/nri803.
Steenland, K., Jin, C., MacNeil, J., Lally, C., Ducatman, A., Vieira, V., Fletcher, T., 2009.
Predictors of PFOA levels in a community surrounding a chemical plant. Environ.
Health Perspect. 117, 1083–1088. https://doi.org/10.1289/ehp.0800294.
Sunderland, E.M., Hu, X.C., Dassuncao, C., Tokranov, A.K., Wagner, C.C., Allen, J.G.,
2019. A review of the pathways of human exposure to poly- and perfluoroalkyl
substances (PFASs) and present understanding of health effects. J. Expo. Sci.
Environ. Epidemiol. 29, 131–147. https://doi.org/10.1038/s41370-018-0094-1.
Tryphonas, H., 2001. Approaches to detecting immunotoxic effects of environmental
contaminants in humans. Environ. Health Perspect. 109, 877–884. https://doi.org/
10.1289/ehp.01109s6877.
United States Department of Health and Human Services-National Toxicology Program
[USDHHS-NTP], 2016. Systematic review of immunotoxicity associated with
exposure to perfluorooctanoic acid (PFOA) or perfluorooctane sulfonate (PFOS)
10
Environment International 156 (2021) 106599
[WWW Document]. accessed 6.8.20. https://ntp.niehs.nih.gov/ntp/ohat/pfoa_pf
os/pfoa_pfosmonograph_508.pdf.
Van Leeuwen, S.P.J., Karrman, A., Van Bavel, B., De Boer, J., Lindstrom, G., 2006.
Struggle for quality in determination of perfluorinated contaminants in
environmental and human samples. Environ. Sci. Technol. 40, 7854–7860. https://
doi.org/10.1021/es061052c.
Watkins, D.J., Josson, J., Elston, B., Bartell, S.M., Shin, H.M., Vieira, V.M., Savitz, D.A.,
Fletcher, T., Wellenius, G.A., 2013. Exposure to perfluoroalkyl acids and markers of
kidney function among children and adolescents living near a chemical plant.
Environ. Health Perspect. 121, 625–630. https://doi.org/10.1289/ehp.1205838.
Yang, Q., Xie, Y., Eriksson, A.M., Nelson, B.D., DePierre, J.W., 2001. Further evidence for
the involvement of inhibition of cell proliferation and development in thymic and
splenic atrophy induced by the peroxisome proliferator perfluoroctanoic acid in
mice. Biochem. Pharmacol. 62, 1133–1140. https://doi.org/10.1016/S0006-2952
(01)00752-3.
Zhu, Y., Qin, X. Di, Zeng, X.W., Paul, G., Morawska, L., Su, M.W., Tsai, C.H., Wang, S.Q.,
Lee, Y.L., Dong, G.H., 2016. Associations of serum perfluoroalkyl acid levels with T-
helper cell-specific cytokines in children: By gender and asthma status. Sci. Total
Environ. 559, 166–173. https://doi.org/10.1016/j.scitotenv.2016.03.187.