University of Southern California Dissertations and Theses

Associations between perfluoroalkyl substances exposure and metabolic pathways in youth

(USC Thesis Other)

Associations between perfluoroalkyl substances exposure and metabolic pathways in youth

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content

Associations between Perfluoroalkyl Substances Exposure and
Metabolic Pathways in Youth

By

Tingyu Yang

A Thesis Presented to the
FACULTY OF THE KECK SCHOOL OF MEDICINE
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
MASTER OF SCIENCE
BIOSTATISTICS

May 2020

Copyright 2020 Tingyu Yang

ii

Acknowledgements
Firstly, I am grateful to my thesis mentor Dr. Zhanghua Chen for her kindness guidance and
strong support of my thesis in the past year. I also want to thank my thesis committee members
Dr. Paul Marjoram and Dr. David V. Conti for their selfless help and guidance.
Additionally, I would like to thank my parents and my friends who gave me great care and
support during this period, which helped me to complete the thesis.
iii

TABLE OF CONTENTS
Acknowledgements ......................................................................................................................... ii
List of Tables ................................................................................................................................. iv
List of Figures ..................................................................................................................................v
Abstract .......................................................................................................................................... vi
1 Introduction ...................................................................................................................................1
2 Method ..........................................................................................................................................2
2.1 Study population ...................................................................................................................2
2.2 Laboratory measures .............................................................................................................3
2.3 Metabolomic features identification .......................................................................................3
2.4 Evaluation of PFASs exposure levels in plasma samples ......................................................4
2.5 Covariate variables .................................................................................................................4
2.6 Statistical analysis...................................................................................................................5
3 Results ...........................................................................................................................................7
4 Discussion ...................................................................................................................................22
References ......................................................................................................................................25

iv

List of Tables
Table 1. Table 1. Sociodemographic characteristics of 102 Meta-AIR youths enrolled during year
2014-2017. .......................................................................................................................................9
Table 2. Plasma concentrations of PFASs exposure presented as geometric means (95%
confidence intervals) among 102 Meta-AIR youths and U.S. population. ....................................10

v

List of Figures
Figure 1. Individual associations between metabolomic features and PFASs exposures among
fasting samples. ..............................................................................................................................11
Figure 2. Linear associations of most significant metabolomic features associated with the
general PFAS exposure. .................................................................................................................12
Figure 3. Significantly HILIC positive metabolic pathways associated with each PFASs
exposures among fasting samples (p-value for mummichog pathway analysis <0.05). ................13
Figure 4. Significantly C18 negative metabolic pathways associated with each PFASs exposures
among fasting samples (p-value for mummichog pathway analysis <0.05). .................................14
Figure 5. Significantly metabolic pathways associated with three PFASs exposures among
fasting samples (p-value for mummichog pathway analysis <0.05). ............................................15
Figure 6. Individual associations between metabolomic features and each PFASs exposures
among 30 minutes post-glucose challenge samples.......................................................................17
Figure 7. Significantly HILIC positive metabolic pathways associated with each PFASs
exposures among 30 minutes post-glucose challenge samples (p-value for mummichog pathway
analysis <0.05). ..............................................................................................................................18
Figure 8. Significantly C18 negative metabolic pathways associated with each PFASs exposures
among 30 minutes post-glucose challenge samples (p-value for mummichog pathway analysis
<0.05). ............................................................................................................................................19
Figure 9. Significantly metabolic pathways associated with three PFASs exposures among 30
minutes post-glucose challenge samples (p-value for mummichog pathway analysis <0.05). .....20

vi

Abstract
Perfluoroalkyl substances (PFASs) are common chemical exposure which have been found to
be ubiquitous among the U.S. population in previous studies. Both animal models in mice and
human studies in adults have discovered that PFASs exposure is related with diabetes. In this study,
we investigate the relationship between metabolic pathways and three PFASs exposures which
include perfluorooctane sulfonate (PFOS), perfluorooctanoic acid (PFOA) and perfluorohexane
sulfonic acid (PFHxS). A total of 102 young adults from Southern California were enrolled in this
study. We used high-resolution metabolomics to assay the concentrations of three PFASs
exposures and examined the individual linear associations between three PFASs plasma
concentrations and metabolic features analyzed by hydrophilic interaction liquid chromatography
(HILIC) positive mode and C18 negative mode high resolution metabolomics using fasting samples
and 30 minutes post-glucose challenge samples. The linear regression models were adjusted for
body fat percent, age, sex, parental education levels, ethnicity, e-cigarette using history, smoking
status in the past 7 days, participation in exercise classes in the last year, self-reported physical
activity level and dietary intake (percent calorie from fat, percent calorie from protein, total calorie
intake and glycemic index). Then we conducted mummichog enrichment pathway analysis with
5000 permutation times to predict the annotations of metabolomic features and explore metabolic
pathways associated with three PFAS chemical exposures. Our results indicated that PFAS
exposures were significantly associated with several lipid and amino acid metabolic pathways
including fatty acid, glycosphingolipids and histidine metabolisms (pathway enrichment test
p<0.05). In conclusion, PFASs can potentially influence dysregulated metabolism in lipids and
amino acids on young adults. Future studies with larger sample sizes and longitudinal measures of
metabolomics are needed to verify our findings. More experimental studies are warranted to
investigate the causal effect of PFAS exposure on dysregulated lipid metabolism.

1

1 Introduction
Perfluoroalkyl substances (PFASs) are a group of chemicals which includes perfluorooctane
sulfonate (PFOS), perfluorooctanoic acid (PFOA), perfluorohexane sulfonic acid (PFHxS) and
many other man-made chemicals. PFASs have been widely used in industry, consumer products
such as fire foams, cooking pans, clothes and utensils during the past decades (Grandjean & Clapp,
2014).
Since PFASs are water-soluble, accumulative and persistent in the environment and have long
half-life in human bodies, PFASs exposure has been detected in the general population. Scientists
have found that PFASs were ubiquitous in the aquatic environment around the world which
suggests that drinking water could be a main source of human exposure to PFASs (Ahrens, 2011).
Additionally, people can be exposed to PFASs through many other pathways including eating
contaminated food and inhalation of indoor dust (Ferguson et al., 2013). A review of Magali Houde
et al. (2006) found that PFASs were detected in the human body all around the world and people
who lived close to urban and industrial areas tended to have higher peripheral levels of PFASs
exposure

(Houde et al., 2006). Kayoko Kato et al. (2011) assessed the exposures to PFASs
including PFOS, PFOA, PFHxS and PFNA in the U.S population and more than 95% of
participants were discovered to have these chemical compounds in their blood samples (Kato et
al., 2011). A positive association between PFASs serum levels and diabetes risks among 7904 men
in the U.S has been shown in a recent cross-sectional study (He et al., 2018). Another cohort study
conducted by Cardenas et al. (2019) also showed that higher PFASs concentration in blood could
contribute to diabetes incidence (Cardenas et al., 2019). Additionally, a longitudinal human study
showed that the exposure to PFASs during childhood could decrease the β-cell function and
contribute to the increased incidence of obesity from age 15 to 21 years old (Domazet et al., 2016).
Moreover, an animal study related to the function of peroxisome proliferator-activated receptor

2

alpha (PPARα) indicated that immunity, cell production and lipid metabolism could be affected
by the exposure to PFOA and PFOS (DeWitt et al., 2009). Results from another animal study
suggested that alterations in several metabolic pathways, such as metabolism of lipids and amino
acids, were associated with PFOA concentration (N. Yu et al., 2016). These studies implied that
metabolomics could help to investigate metabolic mechanisms influenced by PFAS exposures
through findings of altered metabolic pathways associated with PFAS exposures. To date, there
have been few studies on the effect of PFAS exposure on altering metabolic pathways in young
adults who are at the important transition time window from childhood to adulthood. Metabolic
health in young adults may have a long-term impact on their later life. Therefore, this study is
aimed to investigate associations between exposures to PFAS congeners (PFOA, PFOS and
PFHxS) and dysregulated metabolic pathways by profiling metabolome from archived plasma
samples among 102 young adults.
2 Methods
2.1 Study population
A total of 103 young adults (17-22 years old) from Meta-AIR study (Kim et al., 2019) were
included in the data analysis of this study, who were also participants of the Southern California
Children’s Health Study (CHS) (Chen et al., 2015). Participants in this study were obese or
overweight during their high school years (2011-2012). Participants who had any diseases
diagnosis (Type 1 or Type 2 diabetes) or taken any medications were excluded from this study. By
October 2017, we had 103 participants whose plasma samples were measured for high-resolution
untargeted metabolomics in the Clinical Biomarkers Laboratory at Emory University. After the
quality control (QC) method on metabolomics data (more details will be described in the following
paragraphs), we had overall 102 participants included in the final analysis. One additional

3

participant who had lower than 80% detected metabolomic features in the fasting sample was
further excluded in the analysis of fasting untargeted metabolomics data.
2.2 Laboratory measures
All participants took the Oral Glucose Tolerance Test (OGTT). Their plasma samples were
centrifuged for 10 minutes at 2500 revolutions per minute and then stored in sodium heparin 2mL
tubes, which were later examined for insulin level by Human Insulin ELISA Kit (EZHI-14BK).
Additional plasma samples were centrifuged for 15 minutes at 1500 relative centrifugal force and
then stored in sodium fluoride 2mL tubes, which were later examined for glucose level by
hexokinase-mediated reaction on Roche Covas C501. Then all plasma samples were tested for
high-resolution untargeted metabolomics, which was supported by the Children's Health Exposure
Analysis Resource (CHEAR) Program in the Clinical Biomarkers Laboratory at Emory University.
2.3 Metabolomic features identification
Plasma samples collected at fasting and 30 minutes after OGTT were tested for high resolution
metabolomics (HRM) by Fourier transform mass spectrometry (Soltow et al., 2013) and liquid
chromatography. During the process of sample preparation, plasma samples were thawed from -
80 ℃ storage. Then 100 μL ice-cold LC-MS grade acetonitrile was added to 50 μL plasma sample
which was later equilibrated on ice for 30 minutes and centrifuged at 4 ℃ for 10 minutes. Finally,
plasma samples were transferred to 200 μL autosampler vial maintained at 4 ℃ before assay.
Hydrophilic interaction liquid chromatography (HILIC) with positive electrospray ionization
(ESI) and C18 hydrophobic reversed-phase chromatography with negative ESI was used to assay
the plasma samples for metabolomic profiling. The mobile phase flow rate was 0.35 mL/min and
0.5 mL/min during the first 1.5 minutes and the last 4 minutes, respectively. The range of mass-
to-charge ratio (m/z) on the high-resolution mass spectrometer was set from 85 to 1275 and the

4

resolution was 120,000. The method of apLCMS (T. Yu et al., 2009) modified by xMSanalyzer
(Uppal et al., 2013) was applied to extract the raw data. The m/z features in this study represented
distinctive ions which contains ion abundance, m/z and retention time. Additionally, m/z features
with smaller than 10% unidentified values and larger than 100% coefficient of variation were
excluded before data analysis.
2.4 Evaluation of PFASs exposure levels in plasma samples
The LC-HRM method with reverse phase chromatography and negative ESI mode (Go et al.,
2015) was applied to measure the PFOA, PFOS and PFHxS levels in plasma samples and matching
predecessor m/z feature, retention time and MS
2
ion dissociation patterns

to reference standards
was used to verify the compounds. The CHEAR reference blood samples were compared with
NIST standard reference material in 1950 for human plasma (Simon-Manso et al., 2013) to
evaluate the PFOA, PFOS and PFHxS levels. The M-H adduct measured the response factor of
every compound and the approach of single point calibration by using response factors determined
the PFOA, PFOS and PFHxS levels in the plasma samples. The limit of detection (LOD) was
0.1ng/mL for PFOS, 0.02ng/mL for PFOA and 0.03ng/mL for PFHxS. PFASs concentrations were
measured in fasting status samples and 30 minutes post-glucose challenge samples to lessen the
measurement errors and we supposed that oral glucose intake would not affect the PFASs
exposures. After that, the mean concentrations of three PFASs exposures were calculated. In this
study, 100%, 100% and 97.2% of individuals have been identified for PFOS, PFOA and PFHxS,
respectively.
2.5 Covariate variables
In this study, we collected other characteristic data which includes body fat percent, age, sex,
parental education level, ethnicity, e-cigarette using history, smoking status in the past 7 days,

5

participation in exercise classes in the last year, self-reported physical activity level and dietary
intake. Participants were asked whether they have taken any exercise class including swimming,
dancing or gymnastics in the past year and also required to rate their usual physical activity level
on the scale from 0 to 100, which was later categorized into three groups containing less active
level (0-40), moderate active level (50-60) and more active level (70-100). The dietary intake data
collected through two non-consecutive diet sampling in 24 hours (Hoffmann et al., 2002) were
assessed by the Nutrition Data System for Research (version 2014, University of Minnesota).
Participants did the first dietary survey at the study visit and the second dietary survey by phone.
Individuals who answered one of these surveys that the dietary intake was either “less than usual”
or “more than usual” would receive the third survey.
2.6 Statistical analysis
We used the quality control (QC) method to preprocess the metabolomic intensity data,
through which we excluded less than or equal to 50% completeness metabolomic feature data
among all 103 individuals and then excluded less than or equal to 80% completeness individuals
among all metabolomic feature data. Among 103 fasting samples, 6926 out of 8945 HILIC positive
metabolomic features and 6633 out of 9136 C18 negative metabolomic features had higher than
50% completeness. Among 103 post-glucose challenge samples, 6941 HILIC positive and 6619
C18 negative metabolomic features had higher than 50% completeness. Then participants who had
lower than 80% detected metabolomic features were excluded from our analysis (two participants
for fasting samples and one participant for 30 minutes post-glucose challenge samples). Therefore,
our final samples were 6926 HILIC positive and 6633 C18 negative metabolomic features among
101 fasting samples, 6941 HILIC positive and 6619 C18 negative metabolomic features among 102
30 minutes post-glucose challenge samples. After the QC method, we imputed the missing

6

metabolomic feature data with half of the minimum intensities of specific metabolomic feature
and then used quantile normalization method (Lee et al., 2012) to make intensities of metabolomic
features be comparable across subjects, log2 transformation to approximate the skewed
metabolomic feature data to be normal distributed, and pareto scaling approach (van den Berg et
al., 2006) to center the intensity data on the average value and divide the centered data by square
root of the original standard deviations in order to lower the impact of potential outliers.
Next, we applied the linear regressions to investigate the individual associations between
metabolomic features and three PFASs exposures including PFOA, PFOS and PFHxS,
respectively, adjusting for body fat percent, age, gender, parental education level, ethnicity, e-
cigarette using history, smoking status in the past 7 days, participation in exercise classes in the
last year, self-reported physical activity level and dietary intake data (percent calorie from fat,
percent calorie from protein, total calorie intake and glycemic index). Mummichog enrichment
pathway analysis with 5000 permutation times was performed to predict chemical annotations of
metabolomic features and identify metabolic pathways associated with three PFASs chemical
exposures referring to Kyoto Encyclopedia of Genes and Genomes (KEGG) human metabolic
pathway database (Kanehisa & Goto, 2000).
Among fasting samples, we further compared the association results of HILIC positive
metabolomic features across three chemicals and extracted the most significant ones to build the
list for mummichog enrichment pathway analysis for the PFAS exposure (p-value for linear
regression analysis < 0.05). If the same metabolomic feature was significantly associated with two
or more PFASs exposure compounds, we selected the smallest p-value for this metabolomic
feature. We used the same method to analyze C18 negative metabolomic features among fasting

7

samples, HILIC positive metabolomic features and C 18 negative metabolomic features among 30
minutes post-glucose challenge samples, respectively.
3 Results
Among all 102 participants who were 17 – 22 years old, 56.9% and 43.1% of them were males
and females, respectively. The average age was 19.2 years old with one standard deviation equals
to 0.8 years. Most of our participants were Hispanic white (59.8%), had never used e-cigarette
(67.6%) or smoked in the past 7 days (93.1%). 71.6% of participants took some exercise classes
in the last year. For self-reported physical activity status, the number of participants with moderate
activity level was close to that with high activity level. Other detailed sociodemographic
characteristics of participants are shown in table 1.
Table 2 shows the geometric means of PFOA, PFOS and PFHxS exposure levels in plasma
samples among 102 participants were 2.26μg/L (95%CI: 1.61, 3.18), 4.29μg/L (95%CI: 1.61,
11.47) and 1.37μg/L (95%CI: 0.32, 5.79), respectively. The geometric means of three exposures
in this study are comparable to the exposure levels in the general US population during 2013 –
2014 by National Health and Nutrition Examination Survey (NHANES) (CDC).
Among fasting samples, we discovered that 197 metabolomic features were significantly
associated with PFOS, 246 metabolomic features were significantly associated with PFOA and
423 metabolomic features were significantly associated with PFHxS in HILIC positive mode (p-
values for linear regression <0.05) (Figure 1(a), 1(b) and 1(c)). And in C 18 negative mode, 265
metabolomic features were significantly associated with PFOS, 381 metabolomic features were
significantly associated with PFOA and 328 metabolomic features were significantly associated
PFHxS (p-values for linear regression <0.05) (Figure 1(d), 1(e) and 1(f)). After we extracted the
most significant p-values across three PFAS chemical exposures, we found that a total of 742
HILIC positive metabolomic features and 804 C 18 negative metabolomic features were

8

significantly associated with the general PFAS exposure (Figure 2 (a) and (b)). Then we applied
the mummichog enrichment pathway analysis to find metabolic pathways associated with PFASs
exposures.
According to the pathway analysis of HILIC positive features (Figure 3), we discovered that 7
metabolomic pathways were significantly associated with the exposure to PFOS compound, which
contained lipid metabolic pathways including glycerophospholipid metabolism, amino acid
metabolomic pathways including tyrosine metabolism and other altered metabolomic pathways
including vitamin H, B12 and D3, ubiquinone and drug metabolisms. Using mummichog,
exposure to PFOA compound was significantly associated with six altered metabolic pathways,
which contained lipid metabolic pathways including polyunsaturated fatty acid and
glycerophospholipid metabolisms, amino acid metabolomic pathway including histidine
metabolism, degradation metabolic pathways with limonene and pinene metabolisms and other
altered metabolomic pathways including vitamin D3 and B12. Additionally, 6 metabolic pathways
were found to be significantly associated with the PFHxS exposure, which contained lipid
metabolic pathways including fatty acid and linoleate metabolisms and other metabolomic
pathways including purine and pyrimidine metabolisms and vitamin H.

9

Table 1. Sociodemographic characteristics of 102 Meta-AIR youths enrolled during year 2014-
2017.
* These variables are presented as mean (SD) rather than N (%).
† Other races = Asian, African American, Other/Mixed Races.
‡ Current cigarette smoker = smoked in the past 7 days.
‡‡ Exercise Class = took any exercise classes, lessons, or special programs during the past 12 months (outside of
school only).

Sample Size N (%)
Age 19.2 (0.8)*
Sex

Male 58 (56.9)
Female 44 (43.1)
Parental Education

Less than high school 34 (33.3)
Completed high school 35 (34.3)
Some college or higher 26 (25.5)
Unknown 7 (6.9)
Race/Ethnicity

Non-Hispanic White 29 (28.4)
Hispanic White 61 (59.8)
Other
†
12 (11.8)
Ever used e-cigarette
Ever 33 (32.4)
Never 69 (67.6)
Current cigarette smoker
‡

Yes 7 (6.9)
No 95 (93.1)
Participate in exercise class in the last
year
‡‡

Yes 29 (28.4)
No 73 (71.6)
Self-reported physical activity levels

Less active 25 (24.5)
Moderately active 38 (37.3)
More active 39 (38.2)
Dietary variables
Total calorie intake (KJ/day) 1988.8 (627.5)*
Percent calorie from fat 34.3 (8)*
Percent calorie from protein 16.5 (4.9)*
Glycemic index 136.4 (43.5)*

10

Table 2. Plasma concentrations of PFASs exposure presented as geometric means (95%
confidence intervals) among 102 Meta-AIR youths and U.S. population.
PFAS levels (µg/L) All Samples (N=102) U.S. population
†

PFOA
*
2.26 (1.61, 3.18) 1.94 (1.76, 2.14)
PFOS
**
4.29 (1.61, 11.47) 4.99 (4.50, 5.52)
PFHxS
***
1.37 (0.32, 5.79)
1.35 (1.20, 1.52)
* PFOA= perfluorooctanoic acid
** PFOS=perfluorooctane sulfonate
*** PFHxS= perfluorohexane sulfonic acid
† The three exposure levels among U.S. population during 2013 – 2014 were reported by National Health
and Nutrition Examination Survey (NHANES).

According to the pathway analysis of C18 negative features (Figure 4), we found that exposure
to PFOS compound was significantly associated with 12 altered metabolic pathways, which
contained lipid metabolic pathways including glycerophospholipid, linoleate and linoleic acid
metabolisms, amino acid metabolomic pathways including tyrosine metabolism, degradation
metabolic pathways with chondroitin sulfate and heparan sulfate metabolisms and other pathways
including vitamin D3, porphyrin, bile acid and ubiquinone metabolisms. There were 4 metabolic
pathways significantly associated with PFOA exposure, such as the dysfunction of lipid metabolic
pathways including linoleic acid metabolism and other altered metabolomic pathways including
pyrimidine, arachidonic acid metabolisms and vitamin B6. And 3 metabolic pathways were
significantly associated with exposure to PFHxS compound, which contained lipid metabolic
pathways including linoleic acid metabolisms and other metabolomic pathways including drug and
ubiquinone metabolisms.

11

Figure 1. Individual associations between metabolomic features and PFASs exposures among
fasting samples.
Adjusting for body fat percentage, age, gender, parental education level, ethnicity, e-cigarette using history, smoking
status in the past 7 days, participation in exercise class in the last year, self-reported physical activity level and dietary
intake data (percent calorie from fat, percent calorie from protein, total calorie intake and glycemic index), Results for
HILIC positive metabolomic features were shown in (a), (b) and (c) and for C 18 negative metabolomic features were
shown in (d), (e) and (f). X axis represents each metabolite in the function of m/z features and Y axis represents
negative log10 transformed p-value for the linear association. The green dashed line represents p-value = 0.05.

(a) PFOS (b) PFOA (c) PFHxS
(d) PFOS (e) PFOA (f) PFHxS
p=0.05
p=0.05
p=0.05
p=0.05
p=0.05
p=0.05
m/z m/z
m/z
m/z
m/z
-log10(p-value)
-log10(p-value)
-log10(p-value)
m/z
-log10(p-value)
-log10(p-value)
-log10(p-value)
Positive association with exposure (p-value <0.05) Non-significant association with exposure (p-value ≥0.05)
Negative association with exposure (p-value <0.05) p-value=0.05

12

Figure 2. Linear associations of most significant metabolomic features associated with the
general PFAS exposure.
The most significant metabolomic features across three chemicals were extracted to build the list for mummichog
enrichment pathway analysis for the PFAS exposure. If the same metabolomic feature was significantly associated
with two or more PFASs exposure compounds, we selected the smallest p-value for this metabolomic feature. Results
among fasting samples were shown in (a) and (b) and among 30 minutes post-glucose challenge samples were shown
in (c) and (d). X axis represents each metabolite in the function of m/z features and Y axis represents negative log10
transformed p-values. The green dashed line represents p-value = 0.05.

(a) HILIC Positive mode (b) C18 Negative mode
(c) HILIC Positive mode
mmodmodemode
(d) C18 Negative mode
p=0.05
p=0.05
p=0.05
p=0.05
m/z
m/z
-log10(p-value)
-log10(p-value)
m/z m/z
-log10(p-value)
-log10(p-value)
Positive association with exposure (p-value <0.05) Non-significant association with exposure (p-value ≥0.05)
Negative association with exposure (p-value <0.05) p-value=0.05

13

Figure 3. Significantly HILIC positive metabolic pathways associated with each PFASs
exposures among fasting samples (p-value for mummichog pathway analysis <0.05).
The horizontal axis denotes the negative log10 transformed p-values of each pathways and the vertical axis denotes
metabolic pathways. The circle size represents the number of overlapped pathways.
(c) PFHxS
(b) PFOA
(a) PFOS

14

Figure 4. Significantly C18 negative metabolic pathways associated with each PFASs exposures
among fasting samples (p-value for mummichog pathway analysis <0.05).
The horizontal axis denotes the negative log10 transformed p-values of each pathways and the vertical axis denotes
metabolic pathways. The circle size represents the number of overlapped pathways.

(b) PFOA
(c) PFHxS
(a) PFOS

15

Figure 5. Significantly metabolic pathways associated with three PFASs exposures among
fasting samples (p-value for mummichog pathway analysis <0.05).
HILIC positive metabolic pathways was shown in (a) and C 18 negative metabolic pathways was shown in (b). The
horizontal axis denotes the negative log10 transformed p-values of each pathways and the vertical axis denotes
metabolic pathways. The circle size represents the number of overlapped pathways.

(a) HILIC Positive mode
(b) C18 Negative mode

16

Based on the pathway analysis of the extracted metabolomic features significantly associated
with any of three PFASs exposures, 4 metabolomic pathways were detected to be altered by the
PFASs exposures among HILIC positive metabolic features (Figure 5 (a)), containing amino acid
metabolic pathways which included histidine metabolism, lipid metabolic pathways including
fatty acid metabolism and other metabolic pathways including vitamin D3. Among C18 negative
metabolic features (Figure 5 (b)), there were also 4 metabolomic pathways altered by the PFASs
exposures, containing lipid metabolic pathways including linoleic acid metabolism and other
metabolic pathways including ubiquinone, drug metabolisms and vitamin B6.
Among 30 minutes post-glucose challenge samples, we discovered that 342 metabolomic
features were significantly associated with PFOS, 600 metabolomic features were significantly
associated with PFOA and 355 metabolomic features were significantly associated with PFHxS in
HILIC positive mode (p-values for linear regression <0.05) (Figure 6(a), 6(b) and 6(c)). And in
C18 negative mode, 241 metabolomic features were significantly associated with PFOS, 954
metabolomic features were significantly associated with PFOA and 297 metabolomic features
were significantly associated with PFHxS (p-values for linear regression < 0.05) (Figure 6(d), 6(e)
and 6(f)). After we extracted the most significant p-values across three PFAS chemical exposures,
we found that a total of 1088 HILIC positive metabolomic features and 1280 C 18 negative
metabolomic features were significantly associated with the general PFAS exposure (Figure 2 (c)
and (d)).

17

Figure 6. Individual associations between metabolomic features and each PFASs exposures
among 30 minutes post-glucose challenge samples.
Adjusting for body fat percentage, age, gender, parental education level, ethnicity, e-cigarette using history, smoking
status in the past 7 days, participation in exercise class in the last year, self-reported physical activity level and dietary
intake data (percent calorie from fat, percent calorie from protein, total calorie intake and glycemic index), Results for
HILIC positive metabolomic features were shown in (a), (b) and (c) and for C 18 negative metabolomic features were
shown in (d), (e) and (f). X axis represents each metabolite in the function of m/z features and Y axis represents
negative log10 transformed p-value for the linear association. The green dashed line represents p-value = 0.05.

(a) PFOS (b) PFOA (c) PFHxS
(d) PFOS (e) PFOA (f) PFHxS
p=0.05 p=0.05
p=0.05
p=0.05 p=0.05
p=0.05
m/z
m/z m/z
m/z
m/z
-log10(p-value)
-log10(p-value)
-log10(p-value)
-log10(p-value)
-log10(p-value)
-log10(p-value)
m/z
Positive association with exposure (p-value <0.05) Non-significant association with exposure (p-value ≥0.05)
Negative association with exposure (p-value <0.05) p-value=0.05

18

Figure 7. Significantly HILIC positive metabolic pathways associated with each PFASs
exposures among 30 minutes post-glucose challenge samples (p-value for mummichog pathway
analysis <0.05).
The horizontal axis denotes the negative log10 transformed p-values of each pathways and the vertical axis denotes
metabolic pathways. The circle size represents the number of overlapped pathways.

(a) PFOS
(b) PFOA
(c) PFHxS

19

Figure 8. Significantly C18 negative metabolic pathways associated with each PFASs exposures
among 30 minutes post-glucose challenge samples (p-value for mummichog pathway analysis
<0.05).
The horizontal axis denotes the negative log10 transformed p-values of each pathways and the vertical axis denotes
metabolic pathways. The circle size represents the number of overlapped pathways.
(a) PFOS
(b) PFOA
(c) PFHxS

20

Figure 9. Significantly metabolic pathways associated with three PFASs exposures among 30
minutes post-glucose challenge samples (p-value for mummichog pathway analysis <0.05).
HILIC positive metabolic pathways was shown in (a) and C 18 negative metabolic pathways was shown in (b). The
horizontal axis denotes the negative log10 transformed p-values of each pathways and the vertical axis denotes
metabolic pathways. The circle size represents the number of overlapped pathways.

(a) HILIC Positive mode
(b) C18 Negative mode

21

After we applied the mummichog enrichment pathway analysis, in the HILIC positive mode
(Figure 7), we identified that 3 metabolomic pathways were significantly associated with the PFOS
exposure, which contained amino acid metabolomic pathways including alanine and aspartate
metabolisms and lipid metabolic pathways including di-unsaturated fatty acid beta-oxidation and
glycerophospholipid metabolism. PFOA exposure was significantly associated with dysfunction
of 5 metabolic pathways, which contained lipid metabolic pathways including fatty acid and
glycerophospholipid metabolisms, amino acid metabolomic pathway including histidine
metabolism and other altered metabolomic pathways including carnitine shuttle and caffeine
metabolism. Additionally, 7 metabolic pathways were discovered to be significantly associated
with the PFHxS exposure, which contained lipid metabolic pathways including fatty acid
metabolism, amino acid metabolomic pathways including alanine and aspartate metabolisms and
other metabolomic pathways including ubiquinone, carnitine shuttle and purine metabolisms.
According to the analysis of C18 negative metabolomic features (Figure 8), we found that PFOS
exposure was significantly associated with 10 metabolic pathways, which contained pentose
phosphate pathways, lipid metabolic pathways including glycerophospholipid metabolisms,
glycolysis pathways, amino acid metabolomic pathways including tryptophan metabolism,
degradation metabolic pathways including benzoate degradation via CoA ligation and N-Glycan
degradation, other pathways including glycolysis, gluconeogenesis and caffeine metabolisms. And
there were 19 metabolic pathways significantly associated with PFOA exposure, such as the lipid
metabolic pathways including glycerophospholipid, glycosphingolipid and fatty acid metabolisms,
phosphate metabolic pathways and other altered metabolomic pathways including pyrimidine,
vitamin C and aldarate metabolisms. And only 2 metabolic pathways were discovered to be

22

significantly associated with PFHxS compound, containing amino acid metabolic pathways
including tyrosine metabolism and other metabolomic pathways including caffeine metabolism.
Based on the pathway analysis of the extracted metabolomic features significantly associated
with any of three PFASs exposures, 5 metabolomic pathways were detected to be altered by the
PFASs exposures among HILIC positive metabolic features (Figure 9 (a)), containing cholesterol
metabolic pathways, lipid metabolic pathways including glycosphingolipid and fatty acid
metabolisms, amino acid metabolic pathways which included histidine metabolism, and other
metabolic pathways including carnitine shuttle. Among C18 negative metabolic features (Figure 9
(b)), there were also 10 metabolomic pathways altered by the PFASs exposures, containing pentose
phosphate metabolic pathways, lipid metabolic pathways including de novo fatty acid, fatty acid
and glycosphingolipid metabolisms and other metabolic pathways including pyrimidine, ascorbate
and aldarate metabolisms.
4 Discussion
Among fasting samples and 30 minutes post-glucose challenge samples, based on the
information of HILIC positive mode, several lipid and amino acid metabolic pathways were
detected to be significantly associated with each PFASs concentrations, which included fatty acid,
linoleate, glycosphingolipid, histidine and tyrosine metabolisms. There were also some pathways
significantly associated with only one or two of the three PFAS exposures. For instance, among
fasting samples, vitamin H metabolism was significantly associated with both PFOS and PFHxS
concentrations, ubiquinone biosynthesis, limonene and pinene degradation and purine
metabolisms were enriched for PFOS, PFOA and PFHxS, respectively. And among 30 minutes
post-glucose challenge samples, alanine and aspartate, caffeine and purine metabolisms were
significantly associated with PFOS, PFOA and PFHxS, respectively. Some pathways were only

23

associated with only one or two of three PFASs exposures. Bile acid, arachidonic acid and drug
metabolisms among fasting samples and tryptophan, pyrimidine and tyrosine among 30 minutes
post-glucose challenge samples were only enriched for PFOS, PFOA and PFHxS, separately.
Additionally, in the C18 negative mode, mummichog enrichment pathway analysis indicated
that more lipid pathways significantly associated with each of three PFASs exposure levels among
the 30 minutes post-glucose challenge samples comparing to the fasting samples, including de
novo fatty acid, glycerophospholipid, glycosphingolipid and omega-6 fatty acid metabolisms,
which showed the glucose increase may closely related with lipid metabolism. Furthermore, the
results of pathway analysis between individual metabolic features and each PFOS, PFOA and
PFHxS were consistent with the results of extracted features significantly associated with any of
three PFASs exposures.
In this study, we found that several metabolic pathways including lipid and amino acid
metabolism were significantly associated with PFOS, PFOA and PFHxS plasma concentrations
among young adults with obesity or overweight history. These pathways were closely related to
lipid and protein catabolism, energy supply and other cell production. Scientists have found that
some PFASs compounds have the similar chemical structure comparing to the fatty acid
metabolites and may potentially interact the lipid metabolism in human body, which supported our
results from mummichog pathway enrichment analysis (Fletcher et al., 2013). Additionally, our
results were consistent with one previous longitudinal study (Alderete et al., 2019) which
discovered that PFASs exposures could alter the metabolic pathways including fatty acid,
glycosphingolipids, linoleic acid metabolisms among Hispanic children. Our results were also
similar to the study among 965 aged participants (Salihovic et al., 2018) which showed that PFASs

24

exposures were associated with amino acid, fatty acids and glycerophospholipids metabolic
pathways.
One limitation of this study is the relatively small sample size which may lower power and
limit the generalization of our results. Another limitation is the cross-section study design. Results
of our study did not suggest causal effects of PFAS exposures on altered metabolic pathways
including lipid and amino acid metabolisms. Additionally, using raw p-values in the analysis may
induce inflated type I error in the single metabolomic feature findings. While, we used false
discovery rate to adjust for multiple testing after we investigated the individual linear association
between metabolomic features and three PFASs exposures. Due to the small sample size, fewer
than 50 metabolomic signatures were left to be significant with FDR<0.1. Small number of
metabolomic signatures could affect further identifications of potentially important pathways.
Therefore, we used raw p-values with p<0.05 to select metabolomics signatures for pathway
analysis. Nonetheless, the identified metabolic pathways need to be further verified by future
targeted metabolomics studies. Moreover, LC-HRM approach with single point calibration was
used to quantify the concentration of PFASs exposure in plasma samples, which could increase
the cost efficiency and feasibility but have higher measurement error comparing to the traditional
targeted metabolomics analysis with stable isotope-labeled internal standards added to each sample
for specific chemical of interest.
In conclusion, our findings suggested that PFAS exposures can potentially influence
dysregulated metabolism in lipids and amino acids and may have a long-term impact on young
adults. Therefore, efforts are needed to decrease PFAS exposures in order to improve physical
health in young adults.

25

References
Ahrens, L. (2011). Polyfluoroalkyl compounds in the aquatic environment: a review of their
occurrence and fate. Journal of Environmental Monitoring, 13(1), 20–31.
Alderete, T. L., Jin, R., Walker, D. I., Valvi, D., Chen, Z., Jones, D. P., Peng, C., Gilliland, F. D.,
Berhane, K., Conti, D. V., Goran, M. I., & Chatzi, L. (2019). Perfluoroalkyl substances,
metabolomic profiling, and alterations in glucose homeostasis among overweight and obese
Hispanic children: A proof-of-concept analysis. Environment International, 126(January),
445–453. https://doi.org/10.1016/j.envint.2019.02.047
Cardenas, A., Hivert, M.-F., Gold, D. R., Hauser, R., Kleinman, K. P., Lin, P.-I. D., Fleisch, A.
F., Calafat, A. M., Ye, X., & Webster, T. F. (2019). Associations of perfluoroalkyl and
polyfluoroalkyl substances with incident diabetes and microvascular disease. Diabetes
Care, 42(9), 1824–1832.
CDC. Fourth national report on human exposure to environmental chemicals. Available from
https://www.cdc.gov/exposurereport/index.html. Accessed on November 15th 2019.
2019;1-2.
Chen, Z., Salam, M. T., Eckel, S. P., Breton, C. V, & Gilliland, F. D. (2015). Chronic effects of
air pollution on respiratory health in Southern California children: findings from the
Southern California Children’s Health Study. Journal of Thoracic Disease, 7(1), 46.
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. (2009). Immunotoxicity of
perfluorooctanoic acid and perfluorooctane sulfonate and the role of peroxisome
proliferator-activated receptor alpha. Critical Reviews in Toxicology, 39(1), 76–94.
Domazet, S. L., Grøntved, A., Timmermann, A. G., Nielsen, F., & Jensen, T. K. (2016).
Longitudinal associations of exposure to perfluoroalkylated substances in childhood and
adolescence and indicators of adiposity and glucose metabolism 6 and 12 years later: the
European Youth Heart Study. Diabetes Care, 39(10), 1745–1751.
Ferguson, K. K., O’Neill, M. S., & Meeker, J. D. (2013). Environmental contaminant exposures
and preterm birth: a comprehensive review. Journal of Toxicology and Environmental
Health, Part B, 16(2), 69–113.
Fletcher, T., Galloway, T. S., Melzer, D., Holcroft, P., Cipelli, R., Pilling, L. C., Mondal, D.,
Luster, M., & Harries, L. W. (2013). Associations between PFOA, PFOS and changes in the
expression of genes involved in cholesterol metabolism in humans. Environment
International, 57, 2–10.
Go, Y.-M., Walker, D. I., Liang, Y., Uppal, K., Soltow, Q. A., Tran, V., Strobel, F., Quyyumi, A.

26

A., Ziegler, T. R., & Pennell, K. D. (2015). Reference standardization for mass
spectrometry and high-resolution metabolomics applications to exposome research.
Toxicological Sciences, 148(2), 531–543.
Grandjean, P., & Clapp, R. (2014). Changing interpretation of human health risks from
perfluorinated compounds. Public Health Reports, 129(6), 482–485.
He, X., Liu, Y., Xu, B., Gu, L., & Tang, W. (2018). PFOA is associated with diabetes and
metabolic alteration in US men: National Health and Nutrition Examination Survey 2003–
2012. Science of the Total Environment, 625, 566–574.
Hoffmann, K., Boeing, H., Dufour, A., Volatier, J. L., Telman, J., Virtanen, M., Becker, W., &
De Henauw, S. (2002). Estimating the distribution of usual dietary intake by short-term
measurements. European Journal of Clinical Nutrition, 56(2), S53–S62.
Houde, M., Martin, J. W., Letcher, R. J., Solomon, K. R., & Muir, D. C. G. (2006). Biological
monitoring of polyfluoroalkyl substances: a review. Environmental Science & Technology,
40(11), 3463–3473.
Kanehisa, M., & Goto, S. (2000). KEGG: kyoto encyclopedia of genes and genomes. Nucleic
Acids Research, 28(1), 27–30.
Kim, J. S., Chen, Z., Alderete, T. L., Toledo-Corral, C., Lurmann, F., Berhane, K., & Gilliland,
F. D. (2019). Associations of air pollution, obesity and cardiometabolic health in young
adults: The Meta-AIR study. Environment International, 133, 105180.
Lee, J., Park, J., Lim, M., Seong, S. J., Seo, J. J., Park, S. M., Lee, H. W., & Yoon, Y.-R. (2012).
Quantile normalization approach for liquid chromatography–mass spectrometry-based
metabolomic data from healthy human volunteers. Analytical Sciences, 28(8), 801–805.
Salihovic, S., Fall, T., Ganna, A., Broeckling, C. D., Prenni, J. E., Hyötyläinen, T., Kärrman, A.,
Lind, P. M., Ingelsson, E., & Lind, L. (2018). Identification of metabolic profiles associated
with human exposure to perfluoroalkyl substances. Journal of Exposure Science &
Environmental Epidemiology, 1.
Simon-Manso, Y., Lowenthal, M. S., Kilpatrick, L. E., Sampson, M. L., Telu, K. H., Rudnick, P.
A., Mallard, W. G., Bearden, D. W., Schock, T. B., & Tchekhovskoi, D. V. (2013).
Metabolite profiling of a NIST Standard Reference Material for human plasma (SRM
1950): GC-MS, LC-MS, NMR, and clinical laboratory analyses, libraries, and web-based
resources. Analytical Chemistry, 85(24), 11725–11731.
Soltow, Q. A., Strobel, F. H., Mansfield, K. G., Wachtman, L., Park, Y., & Jones, D. P. (2013).
High-performance metabolic profiling with dual chromatography-Fourier-transform mass
spectrometry (DC-FTMS) for study of the exposome. Metabolomics, 9(1), 132–143.

27

Uppal, K., Soltow, Q. A., Strobel, F. H., Pittard, W. S., Gernert, K. M., Yu, T., & Jones, D. P.
(2013). xMSanalyzer: automated pipeline for improved feature detection and downstream
analysis of large-scale, non-targeted metabolomics data. BMC Bioinformatics, 14(1), 15.
van den Berg, R. A., Hoefsloot, H. C. J., Westerhuis, J. A., Smilde, A. K., & van der Werf, M. J.
(2006). Centering, scaling, and transformations: improving the biological information
content of metabolomics data. BMC Genomics, 7(1), 142.
Yu, N., Wei, S., Li, M., Yang, J., Li, K., Jin, L., Xie, Y., Giesy, J. P., Zhang, X., & Yu, H.
(2016). Effects of perfluorooctanoic acid on metabolic profiles in brain and liver of mouse
revealed by a high-throughput targeted metabolomics approach. Scientific Reports, 6,
23963.
Yu, T., Park, Y., Johnson, J. M., & Jones, D. P. (2009). apLCMS—adaptive processing of high-
resolution LC/MS data. Bioinformatics, 25(15), 1930–1936.

Abstract (if available)

Abstract Perfluoroalkyl substances (PFASs) are common chemical exposure which have been found to be ubiquitous among the U.S. population in previous studies. Both animal models in mice and human studies in adults have discovered that PFASs exposure is related with diabetes. In this study, we investigate the relationship between metabolic pathways and three PFASs exposures which include perfluorooctane sulfonate (PFOS), perfluorooctanoic acid (PFOA) and perfluorohexane sulfonic acid (PFHxS). A total of 102 young adults from Southern California were enrolled in this study. We used high-resolution metabolomics to assay the concentrations of three PFASs exposures and examined the individual linear associations between three PFASs plasma concentrations and metabolic features analyzed by hydrophilic interaction liquid chromatography (HILIC) positive mode and C₁₈ negative mode high resolution metabolomics using fasting samples and 30 minutes post-glucose challenge samples. The linear regression models were adjusted for body fat percent, age, sex, parental education levels, ethnicity, e-cigarette using history, smoking status in the past 7 days, participation in exercise classes in the last year, self-reported physical activity level and dietary intake (percent calorie from fat, percent calorie from protein, total calorie intake and glycemic index). Then we conducted mummichog enrichment pathway analysis with 5000 permutation times to predict the annotations of metabolomic features and explore metabolic pathways associated with three PFAS chemical exposures. Our results indicated that PFAS exposures were significantly associated with several lipid and amino acid metabolic pathways including fatty acid, glycosphingolipids and histidine metabolisms (pathway enrichment test p<0.05). In conclusion, PFASs can potentially influence dysregulated metabolism in lipids and amino acids on young adults. Future studies with larger sample sizes and longitudinal measures of metabolomics are needed to verify our findings. More experimental studies are warranted to
investigate the causal effect of PFAS exposure on dysregulated lipid metabolism.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Targeted metabolic signatures and diet in associations with obesity and insulin resistance in young adults

PDF

Linking air pollution to integrative gene and metabolites networks in young adult with asthma

PDF

A hierarchical physiologically-based pharmacokinetic modeling platform for genetic and exposure effects in metabolic pathways

PDF

Modeling mutational signatures in cancer

PDF

Two-step study designs in genetic epidemiology

PDF

Molecular mechanisms of young-onset type 2 diabetes: integration of diet and multi-omics biomarkers

PDF

The role of CD1d transmembrane and cytoplasmic tail domain in CD1d trafficking pathway

PDF

The relationship between per- and polyfluoroalkyl substances, microRNA, and non-alcoholic fatty liver disease

PDF

Using multi-level Bayesian hierarchical model to detect related multiple SNPs within multiple genes to disease risk

PDF

Adaptive metabolic strategies of Mycobacterium tuberculosis to combat stress from antibiotics and ROS

PDF

The association between sun exposure and multiple sclerosis

PDF

Evaluating the associations between the baseline and other exposure variables with the longitudinal trajectory when responses are measured with error

PDF

Cluster detection of burn pits exposure in Iraq and Afghanistan using satellite observations from 2002 to 2012

PDF

The impact of perceived parental and self-reported stress on BMI and body composition in young adults

PDF

Carcinogen metabolism genes, meat intake, and colorectal cancer risk

PDF

Effect of soy isoflavones on anthropometric and metabolic measurements in postmenopausal women

PDF

Association of traffic-related air pollution and age-related macular degeneration in the Los Angeles Latino Eye Study

PDF

Effect of biomass fuel exposure on infant respiratory health outcomes in Bangladesh

PDF

Hierarchical regularized regression for incorporation of external data in high-dimensional models

PDF

Comparative study of the POG and FNCLCC grading systems in non-rhabdomyosarcoma soft tissue sarcomas

Asset Metadata

Creator Yang, Tingyu (author)

Core Title Associations between perfluoroalkyl substances exposure and metabolic pathways in youth

School Keck School of Medicine

Degree Master of Science

Degree Program Biostatistics

Publication Date 05/03/2020

Defense Date 05/15/2020

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag metabolic pathways,metabolomics,OAI-PMH Harvest,perfluoroalkyl substances,Young adults

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Chen, Zhanghua (committee chair), Conti, David (committee member), Marjoram, Paul (committee member)

Creator Email tingyuya@usc.edu,ytingyu@hotmail.com

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c89-295441

Unique identifier UC11663284

Identifier etd-YangTingyu-8404.pdf (filename),usctheses-c89-295441 (legacy record id)

Legacy Identifier etd-YangTingyu-8404.pdf

Dmrecord 295441

Document Type Thesis

Rights Yang, Tingyu

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA