Download as pdf or txt
Download as pdf or txt
You are on page 1of 33

Licensed to Mariana Wollinger Maciel Fassheber - marianawmaciel@gmail.com - 046.925.

309-60

REVIEW ARTICLE
Understanding Environmental Contributions to Autism: Causal
Concepts and the State of Science
Irva Hertz-Picciotto , Rebecca J. Schmidt, and Paula Krakowiak

The complexity of neurodevelopment, the rapidity of early neurogenesis, and over 100 years of research identifying
environmental influences on neurodevelopment serve as backdrop to understanding factors that influence risk and
severity of autism spectrum disorder (ASD). This Keynote Lecture, delivered at the May 2016 annual meeting of the
International Society for Autism Research, describes concepts of causation, outlines the trajectory of research on non-
genetic factors beginning in the 1960s, and briefly reviews the current state of this science. Causal concepts are intro-
duced, including root causes; pitfalls in interpreting time trends as clues to etiologic factors; susceptible time
windows for exposure; and implications of a multi-factorial model of ASD. An historical background presents early
research into the origins of ASD. The epidemiologic literature from the last fifteen years is briefly but critically
reviewed for potential roles of, for example, air pollution, pesticides, plastics, prenatal vitamins, lifestyle and family
factors, and maternal obstetric and metabolic conditions during her pregnancy. Three examples from the case-control
CHildhood Autism Risks from Genes and the Environment Study are probed to illustrate methodological approaches
to central challenges in observational studies: capturing environmental exposure; causal inference when a random-
ized controlled clinical trial is either unethical or infeasible; and the integration of genetic, epigenetic, and environ-
mental influences on development. We conclude with reflections on future directions, including exposomics, new
technologies, the microbiome, gene-by-environment interaction in the era of –omics, and epigenetics as the interface
of those two. As the environment is malleable, this research advances the goal of a productive and fulfilling life for
all children, teen-agers and adults. Autism Res 2018, 0: 000–000. V C 2018 International Society for Autism Research,

Wiley Periodicals, Inc.

Lay Summary: This Keynote Lecture, delivered at the 2016 meeting of the International Society for Autism Research,
discusses evidence from human epidemiologic studies of prenatal factors contributing to autism, such as pesticides,
maternal nutrition and her health. There is no single cause for autism. Examples highlight the features of a high-
quality epidemiology study, and what comprises a compelling case for causation. Emergent research directions hold
promise for identifying potential interventions to reduce disabilities, enhance giftedness, and improve lives of those
with ASD.

Keywords: autism spectrum disorder; environmental risk factors; causal inference; pre- and peri-natal risk factors; pes-
ticides; nutrition; diabetes; gene-environment interaction; epigenetics

Introduction on the mother, despite prominent autism researchers


who published opposing views that drew attention to
Over the last half century, varying conceptual models biologic factors and predicted genetic underpinnings
for etiologic factors in autism spectrum disorder (ASD) [Folstein & Rutter, 1977a; Rimland, 1964]. Twin and
have been proposed, most emphasizing single cause family recurrence studies bolstered the evidence [Fol-
hypotheses. A predominant theory persisting for several stein & Rutter, 1988; Pickles et al., 1995]. By the 1990’s,
decades argued that emotionally unresponsive parent- neuroanatomic observations, neuroimaging data
ing led children to withdraw into their own ‘worlds’ [Courchesne, Yeung-Courchesne, Press, Hesselink, &
and seek comfort in repetitive behaviors [Bettelheim, Jernigan, 1988; Piven et al., 1990], cytogenetics and
1967]. The term “refrigerator mom” laid responsibility links with genetic syndromes [Cohen et al., 1991;

From the Department of Public Health Sciences, MIND Institute (Medical Investigations of Neurodevelopmental Disorders), University of California,
Davis, Davis, California
Grant sponsor: National Institutes of Health; Grant numbers: UG3-OD023365; UL1-TR000002; Grant sponsor: National Institute of Environmental
Health Sciences; Grant numbers: P01-ES011269; P30-ES023513; R01-ES015359; R01-ES020392; R01-ES028089; R01-ES025574; R21-ES021330; T32-
MH073124; Grant sponsor: Eunice Kennedy Shriver National Institute of Child Health and Human Development; Grant number: U54-HD079125;
Grant sponsor: U.S. Environmental Protection Agency; Grant numbers: R829388; R833292; RD-83543201; Grant sponsor: U.S. Department of
Defense; Grant number: AR110194; Grant sponsor: The Allen Foundation.
Received May 24, 2017; accepted for publication October 19, 2017
Address for correspondence and reprints: Irva Hertz-Picciotto, Department of Public Health Sciences, MIND Institute (Medical Investigations of
Neurodevelopmental Disorders), University of California, Davis, Med Sci 1C, Davis, CA, 95616. E-mail: iher@ucdavis.edu
Published online 00 Month 2018 in Wiley Online Library (wileyonlinelibrary.com)
DOI: 10.1002/aur.1938
C 2018 International Society for Autism Research, Wiley Periodicals, Inc.
V

INSAR Autism Research 00: 00–00, 2018 1


Licensed to Mariana Wollinger Maciel Fassheber - marianawmaciel@gmail.com - 046.925.309-60

Folstein & Piven, 1991; Smalley, Tanguay, Smith, & Yamaguchi, 1972]. Those exposed in utero were the
Gutierrez, 1992; Wassink & Piven, 2000; Wassink, most severely affected (by PCBs or mercury) and
Piven, & Patil, 2001] had demonstrated aberrant brain because PCBs are persistent (half-lives of years in
development, establishing a pathobiologic basis for humans), children born to exposed mothers, even years
autism and replacing the psychogenic explanation. By after the incident, suffered similarly [Chen & Hsu,
early 2000s, an intense focus on single gene mutations 1994]. At far lower prenatal levels, PCBs are now known
operating alone or in combination dominated ASD etio- for their ability to disrupt thyroid hormones, immune
logic research [Abrahams & Geschwind, 2008; Bartlett, function, hearing, and neurobehavioral outcomes
Gharani, Millonig, & Brzustowicz, 2005; Bespalova & [Jacobson & Jacobson, 1996; Winneke et al., 1998].
Buxbaum, 2003; Wassink, Brzustowicz, Bartlett, & Szat-
Defining Environment
mari, 2004]. Overwhelming evidence supports a role, in
a subset of persons with ASD, for numerous gene var- Although the word ‘environment’ is often used to refer
iants, with rare ones contributing a small proportion to exogenous chemicals in air, water, soil, food, and
and common variants, as yet largely unidentified, com- household products, our biology and our health are
prising the bulk of the heritability [Gaugler et al., influenced by a much wider array of exposures: nutri-
2014]. Nonetheless, underlying the literature on genet- tion, family resources (financial, educational, social),
ics is an implicit assumption that genes act indepen- household structure, neighborhood attributes (urban/
dently of more mutable factors such as epigenetic suburban/rural, walkability, housing stock, social capi-
marks or environmental exposures, or at least, the latter
tal), workplaces, pets, microbiome, and for the fetus,
can be ignored while the search to understand ASD
the intrauterine environment with maternal health and
genetics advances.
physiology playing a dominant role. What these expo-
Yet the complexity of neurodevelopment, the rapid-
sures share—distinct from inherited nuclear DNA—is
ity of early neurogenesis, and the wide complement of
the capacity to be modified, and this malleability opens
increasing idiopathic developmental disorders with
the door to interventions at various levels. The poten-
challenging disabilities provide a backdrop for a broader
tial to reduce disability by altering these factors is the
perspective on causal factors in ASD. Furthermore, a
premise and foundation of environmental research on
long history of environmental disasters is a compelling
autism.
rationale for interrogating the complex ecology sur-
This paper is divided into four parts. First, we delin-
rounding human development. This tale traces back
eate critical concepts in causation: root causes, suscepti-
more than a century, to use of leaded paints resulting
ble time windows, and multifactorial causation. With
in severe, diverse impairments in young children [Gib-
this foundation, the second section starts with a histori-
son, 1904]. Early studies of exceedingly high exposures
cal overview of environmental etiologic research, then
prompted regulations to reduce lead exposures, and
summarizes the literature on specific prenatal factors
later to the landmark investigation by Needleman
which established, for the first time, that alterations in studied by at least three research groups. Although nei-
behavior and intellectual proficiency occurred at levels ther comprehensive nor systematic, this review briefly
previously considered safe [Needleman et al., 1979]. As but critically evaluates both positive and null findings,
cognitive deficits from lead exposure are not necessarily and where available, meta-analyses. The third section
accompanied by clinical symptoms, they have been selects three examples from the CHildhood Autism
characterized as a ‘silent’ epidemic. High exposure to Risks from Genes and the Environment (CHARGE)
another metal, mercury, from industrial waste in Mini- Study, a large case-control investigation of causes and
mata Bay, Japan [Harada, 1978, 1995] and from epi- contributing factors to ASD, to illustrate methodologic
sodes of grain contamination in Iraq [Harada, 1978, approaches to central challenges in observational (non-
1995] caused severe physical and mental disabilities. As randomized) research: accurate exposure assessment,
with lead, subsequent research identified milder impair- inference from association to causation, and integration
ments associated with low-level mercury exposures usu- of molecular (genetic, epigenetic), with environmental
ally only after adjustment for fish consumption (a influences on health. The fourth and final section out-
major source of both low-level mercury and fatty acids lines future directions for exposomics and gene-
beneficial for brain development) [Karagas et al., 2012; environment interaction.
Oken et al., 2008; Sagiv, Thurston, Bellinger, Amarasiri-
wardena, & Korrick, 2012]. A further example is that of
Concepts of Causation
(polychlorinated biphenyls (PCB’s), an industrial chemi- Root Causes
cal), which mistakenly contaminated cooking oil in two
separate incidents, one in Japan and one in Taiwan A useful classification of causes for any health condi-
[Hsu et al., 1985; Kuratsune, Yoshimura, Matsuzaka, & tion distinguishes root or initiating factors from those

2 Hertz-Picciotto et al./Environmental contributions to ASD INSAR


Licensed to Mariana Wollinger Maciel Fassheber - marianawmaciel@gmail.com - 046.925.309-60

that occur after the disorder or disease is apparent or (CDC) indicate a greater rise in ASD without intellec-
has been diagnosed. These latter exposures may alter tual disability nationally [National Academies of Sci-
course or severity of the disabilities, whereas the root ence, 2015]; these data include more recent years than
causes foster the initial pathogenic changes ultimately the analyses of DDS data, and CDC data sources are
leading to a diagnosable phenotype. Here, we focus on inconsistent over time, that is, cover different regions
exposures, events, or conditions that occur prior to the with widely different prevalence proportions. Unlike
full onset of symptoms, and hence may be causally rele- other data sources, the California DDS system, with its
vant to the emergence of ASD. Symptoms, however, consistent source of case ascertainment throughout sev-
often go unnoticed, delaying diagnoses for years. Some eral decades, provides a robust vehicle for understand-
root causes for ASD may also exacerbate symptoms after ing time trends. Thus, based on two independent
the behavioral syndrome has developed. However, only quantitative analyses [Hertz-Picciotto & Delwiche,
those factors present before symptoms are sufficient to 2009a; King & Bearman, 2009] the rapid, steady
warrant a diagnosis—even if the symptoms went unno- increase in diagnoses in California during the 1990s
ticed—would be considered potentially etiologic. Inter- and early 2000s was only partially attributable to diag-
ventions that reduce or eliminate the root causes would nostic artifacts. An additional analysis showed only 4%
be expected to mitigate the disabling symptoms and of the rise could be attributed to trends toward later
such actions are preventative, in that they lower the maternal age at childbearing [Shelton, Tancredi, &
number of individuals meeting diagnostic criteria. Hertz-Picciotto, 2010]. Paternal age would be expected
Throughout life, however, multiple molecular, physio- to have a similar impact.
logic, microbiologic, chemical, nutritional, family, psy- What explains the rest of the trend? De novo gene
chologic, and other factors may influence autism- mutations and copy number errors of replication occur
related behaviors. These also might serve as targets for too slowly to be plausible explanations. On the other
hand, rapid recent changes in the environment include:
interventions to improve lives of those affected with
increased consumption of sweetened beverages [Bleich
ASD.
et al., 2009]; rise in intake of high fructose corn syrup
Time Trends in Autism from <10g to nearly 100g per person per day [Bray,
Nielsen, & Popkin, 2004], and new ultrasound equip-
In 2003, the California Department of Developmental
ment introduced for fetal imaging with 8-fold higher
Services (DDS), which coordinates services for persons
thermal doses [Webb et al., 2017], both beginning by
with developmental disabilities, including autism,
early 1980’s; increasing C-section deliveries [Osterman
reported a steep and steady rise in cases of autistic dis-
& Martin, 2014]; rising use of brominated flame retard-
order from the start of its electronic records 1990.
ants in household products [Lorber, 2008]; globalization
Notably, the Diagnostic and Statistical Manual (DSM)
of commerce and human travel bringing many new
of mental health conditions was revised to broaden the
viruses across borders; and phthalates becoming perva-
defining symptoms, a change consolidated in the DSM sive in scented products, to name just a few.
IV [American Psychiatric Association, 1994]. A later Parallel trends offer possible clues but, in a society,
analysis of age-specific incidence showed the California marked by rapid changes in all sectors of life, any single
rise to be unrelenting for more than a decade, reaching change is likely confounded by a multiplicity of others.
a 7-fold increase in cumulative incidence to age five Most parallel trends turn out to be coincidences, as
between 1990 and 2001 birth cohorts [Hertz-Picciotto & data mining readily shows (Spurious Correlations
Delwiche, 2009b]. The DSM changes, a trend towards http://tylervigen.com/page?page52). Thus, parallel
earlier age at diagnosis, and a shift in clinical practice trends, by themselves, cannot be construed as evidence
favoring inclusion of milder cases together accounted of causality, and are generally far more likely to be
for only one-third of the observed rise [Hertz-Picciotto chance occurrences.
& Delwiche, 2009b]. Another team reported that about Moreover, causes for autism need not exhibit a rise
a quarter could be attributed to ‘diagnostic substitution’ parallel to that of ASD incidence. Exposures exhibiting
wherein the same behavioral manifestation may have a downward, nonmonotonic, or no trend may still play
prompted a different diagnosis in earlier years [King & a causal role. This apparent paradox could occur if, for
Bearman, 2009]. Given some overlap between diagnos- instance, the chemical of concern was replaced by
tic substitution and younger ages at diagnosis, about another with similar or greater neurodevelopmental
half the rise likely reflected changing clinical practice toxicity. Plasticizers are often replaced with chemicals
rather than a true rise. The California DDS has excluded having very similar structures. More generally, the net
cases without major impairments, and beginning in impact of multiple risk factors with changing prevalen-
2003, cases with fewer than three functional limita- ce’s may negate the impact of trends in any particular
tions. Data collected by the Centers for Disease Control causal factor.

INSAR Hertz-Picciotto et al./Environmental contributions to ASD 3


Licensed to Mariana Wollinger Maciel Fassheber - marianawmaciel@gmail.com - 046.925.309-60

Finally, the other ‘half’ of the rise in autism, that is, close to conception, but not in later time periods,
the previously unrecognized cases now being diag- according to two separate studies [Schmidt et al., 2011,
nosed, as well as the background of cases that have 2012; Suren et al., 2013]. Studies on maternal fever sug-
been diagnosed all along, also require etiologic explana- gest first and second trimesters as being influential
tion. Both genetic susceptibility and environmental fac- [Atladottir et al., 2010], whereas air pollution results
tors may be involved. indicate the strongest associations with ASD following
third trimester or early postnatal exposures. A broad
Critical Time Windows
survey of parents regarding stressors during pregnancy
Another key concept for understanding causation is implicated weeks 21–32 in autism [Beversdorf et al.,
that of the ‘critical time window’. Autistic symptoms 2005]. Thus, time interval(s) for brain vulnerability may
commonly appear in the second and sometimes the vary by the agent, its mechanism, and the target of its
first year of life, indicating a strong likelihood of origins action. A critical window may be identifiable for expo-
in the prenatal and early postnatal periods. Neurogene- sures that are uncorrelated across time periods (e.g.,
sis takes place at an astonishing rate, averaging 250,000 fever) or for associations not observed in other time
new neurons per minute during gestation; the result is windows (prenatal vitamins, thalidomide), but difficult
100 billion neurons at birth [Davis & Pfaff, 2014]. Most to discern when exposures are highly correlated over
of the growth is in the third trimester, when 40,000 time.
synapses are formed per minute. Moreover, brain devel- Often the critical, hence relevant, window for a par-
opment comprises an array of qualitatively different ticular exposure is unknown. Or, the exposure informa-
processes with overlapping timing: neural tube forma- tion is only available in broad intervals enveloping
tion, cell proliferation and differentiation, migration, periods outside the susceptible window. In these situa-
dendritic arborization, synaptogenesis, apoptosis, for- tions, relevant exposure may be measured with error,
mation, and connectivity of cortical mini-columns, and and if exposure is categorized, will result in misclassifi-
myelination. Synaptogenesis, considered to play a cation [Hertz-Picciotto, Pastore, & Beaumont, 1996].
major role in ASD, accelerates in mid-2nd trimester and Errors in capturing exposure in the susceptible window
continues through adolescence, whereas neural migra- will usually be independent of the diagnosis, that is,
tion is completed by birth [Tau & Peterson, 2010]. just as likely in those with, as in those without, ASD
Evidence from ASD genetics implicates disruption of (‘nondifferential misclassification’). These errors bias
these processes. Variants enriched in ASD are found in the association towards the null, that is, towards find-
genes that regulate synaptogenesis [De Rubeis et al., ing no effect, when both exposure and outcome are
2014; Patel, Ioannidis, Cullen, & Rehkopf, 2015], cell dichotomous variables [Rothman, Greenland, & Lash,
migration [Boitard et al., 2015; Reiner, Karzbrun, Kshir- 2008]. Consequently, lack of knowledge about the criti-
sagar, & Kaibuchi, 2016; Wang et al., 2014; Wong cal window for a given exposure can lead to Type 2
et al., 2014], dendritic growth [De Rubeis et al., 2014] errors in statistical tests (null hypothesis not rejected
as well as more basic functions such as chromatin when the alternative hypothesis is true), and underesti-
remodeling, cell signaling and cell cycle regulation mation of effect sizes.
[Anney et al., 2011; De Rubeis et al., 2014]. Transcrip- Multifactorial Causation
tomics along with whole exome sequencing of histone
modifying genes similarly suggest altered expression of Much of biomedical science proceeds by tackling single
genes involved in synapse formation, neuronal connec- hypothesized causes of medical conditions, even when
tivity, and apoptosis [De Rubeis et al., 2014; Mahfouz, multiple risk factors are established. Although the con-
Ziats, Rennert, Lelieveldt, & Reinders, 2015; Zeidan- cept that a single or small group of genes would explain
Chulia et al., 2014; Zhang et al., 2016]. Gene expression the vast majority of ASD cases had considerable traction
is regulated by epigenetic changes, which in turn may not long ago, today, it is broadly acknowledged that
have environmental origins. hundreds of genes may contribute to ASD risk. Thus, a
Brain development unfolds with precise timing and prevailing consensus accepts multifactorial causation of
spatial attributes, and several epidemiologic studies sug- ASD across the population.
gest specificity of susceptible windows for environmen- Less appreciated is the multifactorial causation within
tal influences on ASD, summarized in [Schmidt, Lyall, an individual. The impacts of causal factors—both
& Hertz-Picciotto, 2014a]. Empirical research on thalid- genetic and environmental—are often interdependent,
omide traced the exposures to an early insult on days and a person’s inherent ability to adapt will determine
20–24 days post-conception [Rodier, Ingram, Tisdale, their ‘tipping point.’ Whereas each of us inherits
Nelson, & Romano, 1996]. Maternal intake of prenatal genetic susceptibility, along with components of our
vitamin supplements appeared protective when taken parental epigenetic profiles, our environmental ‘toxic

4 Hertz-Picciotto et al./Environmental contributions to ASD INSAR


Licensed to Mariana Wollinger Maciel Fassheber - marianawmaciel@gmail.com - 046.925.309-60

experience, even in the earliest stages of life, it is not neces-


sary to identify and remove or reduce them all in order to
lessen the risk! Instead, large gains in improving the lives
for those at risk can be achieved by focusing on com-
monly encountered factors that are disabling for large
proportions of people. In other words, some of the
greatest impacts will occur when interventions success-
fully target the most prevalent causal factors.

Etiologic Research on Environmental


Contributions to ASD: A Developmental
Perspective
The 20th Century
Pre-conception of environmental etiologic research.
Figure 1. Causal ‘pies’. This pie represents a hypothetical set
Figure 2 provides a timeline of ASD etiologic research.
of causes that all together is minimally sufficient to induce the
At the turn of the 21st century, evidence for genetic
disease or condition of concern in some members of the popula-
tion. Represented here is one possible pie, or set of sufficient contributions to ASD comprised high family recurrence,
causes, but there may be (usually are) many different sufficient twin concordance, chromosomal abnormalities and
sets across the population. For any such set, removal of any some identified candidate genes [Bailey et al., 1995; Fol-
one factor will make the set insufficient, and therefore the indi- stein & Rutter, 1977b,; Ritvo et al., 1989; Wassink &
vidual would not meet criteria for the diagnosis, though lesser Piven, 2000; Wassink et al., 2001]. At the same time,
impairments might still occur. understanding of nongenetic root causes for ASD
amounted to a few clues about infectious origins and
burden’ or ‘load’ also influences risk. As described by pharmacologic agents. These included an unusually
Herbert [Herbert & Weintraub, 2012] the toxic load high prevalence of autistic symptoms in children
includes chemical, infectious, psychologic, nutritional, exposed to congenital rubella [Chess, 1977; Desmond
and other stressors, and can erode an individual’s resil- et al., 1967], prenatal thalidomide, a medication pre-
ience through effects on the epigenome, endocrine scribed as a sedative for morning sickness in pregnancy
homeostasis, cell signaling and a range of physiologic [Rodier et al., 1996; Stromland, Nordin, Miller, Aker-
and molecular adaptive mechanisms, ultimately over- strom, & Gillberg, 1994], or prenatal valproate expo-
whelming adaptive capacity. From the perspective of a sures [Rodier et al., 1996]. Another epidemiologic study
multifactorial model encompassing both genetic and suggested that maternal gestational or infant (first 18
environmental factors, genetic inheritance determines postnatal months) influenza, ear infection, or fever of
baseline susceptibility, and epigenetics—both inherited unknown origin, occurred more often in children with
and acquired via interaction with the environment— autism than in their unaffected siblings [Deykin & Mac-
likely acts by setting or resetting that threshold for Mahon, 1979]. Other studies indicated links between
adaptation. ASD and suboptimal birth outcomes (preterm delivery,
One conceptualization of multifactorial causation is APGAR score) [Nelson, 1991], which may represent
presented in Figure 1. This ‘pie’ represents a minimally causal agents, mediating factors, or markers of little eti-
sufficient set of causes, that is, together, the factors in a ologic significance. Overall, research on risk factors or
single pie will result in ASD in at least some segment of causes of autism was scattered: as a field of inquiry, this
the population [Rothman et al., 2008]. These multiple was the pre-conception period.
causes need not occur simultaneously. For instance,
The 21st Century
vitamins might be influential around conception,
maternal illnesses in mid- to late gestation, and certain Conception & gestation of environmental etiologic
exogenous chemicals even later or earlier. Across the research. Continuing on our timeline, despite a
population, there may be dozens or hundreds of such largely barren landscape of nongenetic research at the
pies. turn of the 21st century, a new wave of environmental
Under this conceptual model of multifactorial causa- studies began to appear. Initially, many were ‘ecologic’
tion, removal of one piece of the pie results in a set studies, denoting a particularly weak epidemiologic
that is no longer sufficient. The resulting phenotype design that correlates group-level outcomes such as
might range from virtually free of symptoms to having prevalences or rates, with a group-level statistic for
clear impairments but not meeting diagnostic criteria. exposure (median levels, proportion exposed, etc.).
Thus, despite thousands of stressors in the human Such studies are prone to various biases, including

INSAR Hertz-Picciotto et al./Environmental contributions to ASD 5


6
Hertz-Picciotto et al./Environmental contributions to ASD
Figure 2. Timeline of etiologic research on autism. Shown is a broad sweep of research on autism etiology, highlighting major framing publications on autism incidence, prev-
alence and causation (white background), along with genetics (black background), and emphasizing prenatal modifiable factors: medical (medium blue background, environmen-
tal chemical (grey background), and gene-by-environment interactions (light blue background).
Licensed to Mariana Wollinger Maciel Fassheber - marianawmaciel@gmail.com - 046.925.309-60

INSAR
Licensed to Mariana Wollinger Maciel Fassheber - marianawmaciel@gmail.com - 046.925.309-60

confounding at the group-level, reverse causation using the nationally representative NHANES (National
(when exposure and outcome are measured in the same Health and Nutrition Examination Survey) provided
time interval, the outcome could be influencing the unequivocal evidence that household income is posi-
exposure), and cross-level confounding (the latter tively correlated with numerous serum nutrients, and
referred to as “ecologic” bias). [Morgenstern, 1995]. negatively correlated with blood and urine measure-
ments of metals, hydrocarbons and volatile organics
that are typical constituents of ambient air pollution,
Birth. In 2006–2007, high quality individual-level
along with a wide range of viral and inflammatory
studies emerged. Within a decade, scores of reports pro-
markers [Patel et al., 2015]. Notably, nutritional, viral
duced results on environmental chemical and physical
and inflammatory exposures have each been associated
agents, lifestyle, infectious agents, trauma, and medical
with ASD. Moreover, two large rigorous studies con-
conditions or interventions. We review major classes of
ducted in Europe, where there is less of a link between
risk or protective factors occurring in the prenatal
wealth and clean air, showed no association with ASD
period in Tables 1, 2, and 3. This literature, represents
[Gong et al., 2017; Guxens et al., 2016]. Additionally, a
the Infancy & Childhood of the field. A separate sec- troubling discrepancy is the inconclusive literature on
tion probes, more deeply, methodologic issues relevant cigarette smoking in relation to ASD, even though
to causal inference using three instructive examples of tobacco smoke and ambient air pollution contain thou-
environmental chemicals, nutrition, and maternal med- sands of chemicals in common, some associated with
ical conditions, with special reference to work con- ASD in epidemiologic studies and many with biologic
ducted at the UC Davis MIND Institute. plausibility. Moreover, many studies on tobacco smoke
Chemicals and Pollutants (Table 1) failed to control for confounding or inappropriately
controlled for potential intermediates (such as birth-
Air pollution. Over a dozen full-scale epidemiologic weight). In short, unanswered questions about con-
studies have been published, most indicating that pre- founding in studies of air pollution and about the true
natal exposure to air pollutants increased risks for ASD association of smoking with ASD remain.
[Becerra, Wilhelm, Olsen, Cockburn, & Ritz, 2013;
Hartz, Heinonen, Shapiro, Siskind, & Slone, 1975; Kalk-
brenner et al., 2010; Kalkbrenner, Schmidt, & Penlesky, Pesticides. Studies of pesticides and ASD are dis-
2014; Raz et al., 2015; Talbott et al., 2015; Volk, Hertz- cussed under “Methodologic Challenges” below.
Picciotto, Delwiche, Lurmann, & McConnell, 2011;
Volk, Lurmann, Penfold, Hertz-Picciotto, & McConnell, Phthalates. Phthalates are a ubiquitous class of
2012; Windham, Zhang, Gunier, Croen, & Grether, chemicals used as plasticizers, stabilizers, emulsifiers,
2006]. However, different pollutants were evaluated, dispersants, solvents, and lubricants and as enteric coat-
including coarse and fine particles, nitrogen dioxide, ings on pharmaceuticals and nutritional supplements.
ozone, metals, solvents, and diesel particles. Metrics of They are in vinyl flooring, adhesives, pesticides, food
exposure varied from proximity to freeways to model- packaging, children’s toys, medical tubing, air fresh-
based estimates of concentrations, sometimes based on eners, toiletries (soaps, shampoos, hair products), and
annual averages, and in other studies based on more cosmetics, [Centers for Disease Control and Prevention,
temporally sensitive models. The models are con- 2016; Kim, Hong, Bong, & Cho, 2015; U.S. Environ-
structed from air monitoring, source information, dis- mental Protection Agency Office of Prevention, 2006].
persion models, and sometimes also land topography In short, they are ubiquitous. Four studies have
and meteorologic variables; model validation was per- addressed phthalates in relation to ASD, autistic symp-
formed for many, though for limited time periods and toms, or common co-occurring behaviors. In a Scandi-
geographic areas. Outcomes also varied from maternal navian population-based sample, maternal report of a
response to a single question about whether they have child with ASD was associated with vinyl flooring in
a child with autism to a full clinical evaluation using the bedrooms [Larsson, Weiss, Janson, Sundell, & Bor-
established standardized instruments (e.g., ADOS and nehag, 2009]. Phthalates are a major component in
ADI-R). vinyl flooring and off-gas continuously. Two cohort
The pattern of findings is generally consistent across studies measured phthalate metabolites in maternal 3rd
U.S. studies, though not necessarily for the same pollu- trimester urine samples: one found poorer scores on all
tants. Although virtually all studies adjusted for at least domains of the Social Responsiveness Scales (SRS) [Mio-
one socioeconomic variable, concerns about residual dovnik et al., 2011], whereas the other measured differ-
confounding remain: Air pollution is highest in com- ent phthalates, and showed no association with the
munities with the lowest socioeconomic levels in most composite SRS [Braun et al., 2014b]. In a fourth study,
of the U.S., and, a recent analysis of correlates for SES phthalates measured in house dust samples collected

INSAR Hertz-Picciotto et al./Environmental contributions to ASD 7


Licensed to Mariana Wollinger Maciel Fassheber - marianawmaciel@gmail.com - 046.925.309-60

Table 1. Evidence on Environmental Chemicals and Pollutants in Relation to Autism Spectrum Disorder or Autism Symptoms
Environmental chemical
or pollutant class: State of the evidence: Reviews or influential references:

Air pollution Associated with increased risk of autism in close to a dozen studies in the Windham et al. [2006]; Kalkbrenner et al.
U.S. and Asia, but not in several European populations. Different exposure [2010]; Volk et al. [2011, 2012]; Roberts
measures have included: (a) modeled estimates by Census tract; (b) prox- et al. [2013]; von Ehrenstein et al.
imity of geocoded home address to freeway; (c) model-based estimates of [2014]; Raz et al. [2015]; Guxens et al.
NO2, ozone, PM2.5, and PM10 at geocoded home address; (d) other air con- [2016]; Gong et al. [2017]
taminant monitoring and land use regression. Results most consistent for
PM2.5; other pollutants not entirely consistent. Two European studies do
not confirm air pollution associations with ASD. Because of strong correla-
tions between air pollution and socioeconomic status, residual confound-
ing may be affecting multiple studies. Apparent discrepancy with cigarette
smoke (see below) also raises questions. Air pollutants induce oxidative
stress and have been associated with cognitive and neurobehavioral
deficits.
Pesticides: Organophos- Four studies provide evidence, all used objective measures of exposure. Two Rauh et al. [2011]; Eskenazi et al. [2007];
phates (OP’s) cohort studies reported prenatal OP exposures associated with PDD symp- Roberts et al. [2007]; Shelton et al.
toms: one measured chlorpyrifos in plasma (cord or maternal); the other [2014]
measured metabolites (nonspecific for organophosphate pesticides) in
urine. Two case-control studies linked proximity to agricultural applica-
tions of OP’s with autism risk. Third trimester exposures to chorpyrifos,
the most commonly used OP, showed OR>3.0. Pesticide drift measured in
air samplers at distances several kilometers away from agricultural applica-
tions correlates with amounts applied [Wofford et al., 2014]. Organophos-
phate exposures during gestation have also been associated with
cognitive impairment, volumetric differences in brain regions, attention
deficits, and tremor (see Table 4).
Pesticides: Pyrethroids Association observed with applications of pyrethroids in two studies, includ- Roberts et al. [2007]; Shelton et al. [2014]
ing two large case-control studies. Both geocoded residential addresses
and assessed proximity to agricultural applications during the pregnancy.
One observed association with bifenthrin, which degrades most slowly of
all pyrethroids. The other study found total pyrethroids or Type II pyreth-
roids associated primarily when exposures were in the pre-conception or
3rd trimester, after adjustment for confounders. Numerous mechanisms
have been proposed for neurodevelopmental toxicity [Shelton et al.,
2012].
Phthalates Three of four studies suggest an association with neurobehavior; one of two Larsson et al. [2009]; Miodovnik et al.
reported an association with ASD diagnosis, and one of two found mater- [2011]; Braun et al. [2014a]; Philippat
nal prenatal urinary phthalate metabolites associated with higher SRS et al. [2015]
scores. One observed greater hyperactivity/impulsivity and inattention.
Studies used: (a) presence of vinyl flooring; (b) phthalate metabolites
measured in maternal prenatal urine; and (c) house dust phthalates. Given
the continuous nature of most phthalate exposure sources, urinary meas-
urements may provide reasonable approximations to chronic long-term
exposures. Phthalates disrupt androgen activity [Main et al., 2006;
National Research Council, 2008; Miodovnik et al., 2014].
Persistent pollutants: Three studies of PCBs in association with ASD or SRS scores, one with very Cheslack-Postova et al. [2013]; Braun et al.
PCBs low power. Two others (one using SRS, one using ASD diagnosis) found [2014a]; Lyall et al. [2017]; Jolous-
associations of higher scores or risk for the same congener peak (IUPAC Jamshidi et al. [2010]
#138/#158), while one found lower SRS and the other found higher ASD
risk with #153. Nondioxin-like PCBs can influence neuronal dendritic
development. Toxicologic studies demonstrate alterations in social behav-
iors in rodents. Disruption of thyroid hormone homeostasis is a mecha-
nism of neurodevelopmental toxicity in rodents.

ASD, autism spectrum disorder; PCBs, polychlorinated biphenyls; PDD, pervasive developmental delay.

after the child was diagnosed were associated with to obtain deep dust deposited over a period of years
other developmental delay but not autism; however, in that included, but was not specific to, prenatal expo-
both groups, higher phthalate levels were associated sures. Thus, three of four studies in different popula-
with hyperactivity-impulsivity and inattention [Philip- tions, using three different methods to assess early life
pat et al., 2015]. The house dust protocol was designed exposure suggest potential impact on neurobehavior,

8 Hertz-Picciotto et al./Environmental contributions to ASD INSAR


Licensed to Mariana Wollinger Maciel Fassheber - marianawmaciel@gmail.com - 046.925.309-60

but not necessarily ASD. Phthalates have well-known latter now contaminating drinking water in the U.S.
anti-androgenic properties [National Research Council, [Hoffman et al., 2011; Post, Louis, Lippincott, & Proco-
2008; Main et al., 2006; Miodovnik, Edwards, Bellinger, pio, 2013]. Both PBDEs and PFASs are endocrine-
& Hauser, 2014] and sexual dimorphism is commonly disrupting and have been linked with neurologic and
observed in epidemiologic studies of phthalates and cer- neurodevelopmental outcomes in a small number of
tain behavioral and cognitive outcomes [Engel et al., human studies [Braun et al., 2017; Herbstman et al.,
2010; Whyatt et al., 2012]. With a high male:female 2010; Hoffman, Webster, Weisskopf, Weinberg, &
ratio in autism, an ‘extreme male brain’ resulting from Vieira, 2010; Lopez-Espinosa, Mondal, Armstrong, Eske-
excessive androgens during gestation has been proposed nazi, & Fletcher, 2016; Sagiv et al., 2015; Stein & Savitz,
[Baron-Cohen et al., 2015]. The literature, however, is 2011; Zhao et al., 2015].
contradictory [e.g., Kung et al., 2016]. Alternative more
Proxy Exposures
biologically focused frameworks emphasize sexually
dimorphic fetal programming of stress responses by the Maternal and paternal age, inter-pregnancy interval (see
HPA-placental axis [Davis & Pfaff, 2014] and brain mas- below) and season are all factors that can serve as surro-
culinization in context of immunoregulatory processes gates for more proximal risk or protective factors. For
[McCarthy & Wright, 2017]. example, season of conception or birth can be a proxy
for cold/warm weather, air pollution, influenza preva-
lence, use of pesticides, or availability of fruits and veg-
Persistent pollutants. PCBs are a class of chemicals
etables. Winter conceptions have been associated with
formerly used in industrial applications but now
higher incidence of ASD [Gadow, Perlman, Ramdhany,
banned; by 1985, production ended worldwide. They
& de Ruiter, 2016; Mackay et al., 2016; Zerbo, Iosif, Del-
are persistent in the environment bioaccumulate
wiche, Walker, & Hertz-Picciotto, 2011], though other
through the food chain, have been associated with neu-
studies found different seasons [Atladottir et al., 2007;
rodevelopmental deficits in young children [Vreugden-
Hebert et al., 2010; Hultman et al., 2002; Lee et al.,
hil, Lanting, Mulder, Boersma, & Weisglas-Kuperus,
2008]. As the net effect of season is likely not strong,
2002; Winneke et al., 1998] and as recently as 2003–
only large studies would have sufficient power.
2004 were still detected in 100% of the U.S. population
[Megson et al., 2013]. Three studies have investigated Lifestyle Factors (Table 2)
associations of PCBs with ASD or with SRS scores. One
General nutrition. Evidence is accumulating for a
small study reported elevated OR’s but confidence inter-
potentially large role in autism etiology for gestational
vals (CI) were very wide [Cheslack-Postava et al., 2013].
nutrition. Several studies examining maternal nutrients
In a cohort from Cincinnati, one specific PCB peak
or diet support an association with folic acid and other
(138/158) was associated with higher composite SRS
nutrients [Lyall, Schmidt, & Hertz-Picciotto, 2014;
scores in a semi-Bayesian analysis with 52 pollutants,
Schmidt et al., 2011, 2012; Suren et al., 2013]; folic acid
but this may have been driven by a very high correla-
studies are reviewed in detail below (‘Methodologic
tion with another PCB (153) showing the opposite asso-
Challenges’). Here we discuss iron, fatty acids, and fish
ciation [Braun et al., 2014b]. However, a large nested
consumption, and short inter-pregnancy interval (IPI)
case-control study that measured maternal PCBs in 2nd
as a proxy. We omit postnatal factors (e.g., breastfeed-
trimester serum and confirmed diagnoses based on an
ing) due to space constraints.
expert review of medical records, found that these same
two PCB peaks (138/158 and 153) were both associated
with significantly elevated risk for ASD [Lyall et al., IPI. As recently systematically reviewed, a short IPI
2017b]. PCBs can disrupt thyroid hormones, though has been consistently associated with ASD, conferring
effects appear small at today’s population levels. about 2-fold higher risk for children born after an IPI
Other persistent pollutants that are pervasive in less than 12 months compared with those born after 36
homes and detectable in the vast majority of U.S. resi- or more months [Cheslack-Postava, Liu, & Bearman,
dents and in other populations have yet to be ade- 2011; Cheslack-Postava et al., 2014; Conde-Agudelo,
quately studied as root causes of autism. Two examples Rosas-Bermudez, & Norton, 2016; Coo et al., 2015; Dur-
are polybrominated diphenyl ethers (PBDEs), a class of kin, DuBois, & Maenner, 2015; Gunnes et al., 2013;
compounds used as flame retardants in a wide variety Zerbo, Yoshida, Gunderson, Dorward, & Croen, 2015].
of household and building products [Wu et al., 2015] Short IPI may indicate maternal depletion of nutrients,
and perfluoroalkyl and polyfluoroalkyl substances such as folate, as essential nutrients are preferentially
(PFAs), which are ubiquitous due to their use as stain distributed to the developing fetus during pregnancy,
resistant coatings on fabrics and carpets, as nonstick and can remain low for up to a year postpartum
coatings on cookware, and in firefighting foam, the [O’Rourke, Redlinger, & Waller, 2000; Smits & Essed

INSAR Hertz-Picciotto et al./Environmental contributions to ASD 9


Licensed to Mariana Wollinger Maciel Fassheber - marianawmaciel@gmail.com - 046.925.309-60

Table 2. Evidence on Lifestyle Factors in Relation to Autism Spectrum Disorder or Autism Symptoms
Lifestyle & Diet State of Evidence Reviews and Influential References2

Interpregnancy interval Shorter intervals (<12 months) from end of the previous pregnancy to the Cheslak-Postava et al. [2011]; Dodds et al.
start of pregnancy associated with about two-fold higher autism risk in [2011]; Gunnes et al. [2013]; Cheslack-
8 studies. Some evidence for dose-response, however three studies also Postava et al. [2014]; Coo et al. [2015];
showed higher risk with longer intervals. Associations could result from Durkin et al. [2015]; Zerbo et al. [2015];
maternal nutrient depletion or cultural, family planning, or lifestyle Conde-Agudelo et al. [2016]
factors.
Maternal folic acid and Higher folic acid intake and folic acid supplements associated with reduced Schmidt et al. [2011, 2012, 2017]; Suren
prenatal vitamins risk for ASD or autistic traits in several population-based studies; one et al. [2013]; Braun et al. [2014a];
study had null findings. Association with ASD diagnosis specific to expo- Steenweg-de Graaff et al. [2015]; Virk
sure near conception in two studies. Evidence for a stronger association et al. [2016]
when mother or child has genetic or environmental susceptibility. Two
studies observed no association of autistic traits with maternal serum
folate collected after the first trimester of pregnancy, similar to reported
lack of association with folic acid intake or supplements after first two
months. Mechanisms for folic acid protection may involve methylation of
DNA and altered gene expression. One study of interactions showed folic
acid protection against pesticide-related increased risk of ASD.
Maternal iron In a large case-control study, mothers of children with ASD were less likely Schmidt et al. [2014b]; Suren et al. [2013]
to report taking an iron supplement; lowest quintile of supplemental iron
associated with ASD, especially in older mothers and mothers with meta-
bolic conditions. A large birth cohort found no association with minerals.
Maternal fatty acids and Literature is inconsistent. Exposures have included (a) prenatal fish oil sup- Suren et al. [2013]; Lyall et al. [2013];
fish consumption plements, (b) intake of polyunsaturated fatty acids, omega-6 and omega-3 Steenweg-de Graaff et al. [2016]; Julvez
fatty acids and/or linoleic acid, and (c) fish consumption. Self-reported et al. [2016]
intake of polyunsaturated fatty acids, omega-6 fatty acids and linoleic
acid associated with lower ASD risk in one study, but in another, total
omega-6 and linoleic acid associated with more ASD traits. Another
observed increased ASD risk with very low omega-3 intake. Fish intake not
associated with ASD or traits in two studies, while another found fewer
autistic traits in association with high fatty fish consumption; this latter
was the only study to adjust for mercury.
Maternal smoking Evidence is inconsistent. Two recent meta-analyses, each based on fifteen Rosen et al. [2015]; Tang et al. [2015].
studies, reported the same OR of 1.02 (95% CI 5 0.93, 1.12 or 1.13).
However, two-thirds of original analyses either did not adjust for any con-
founders, or incorrectly adjusted for birthweight or gestational age,
potential intermediates, for which adjustment can induce bias. All relied
on self-report of smoking, known to be prone to underascertainment. No
studies examined genetic susceptibility, nor have objective exposure
markers been utilized in smoking assessment.

2001; van Eijsden, Smits, van der Wal, & Bonsel, 2008]. decreased social interaction [Pollitt, 1993]. Iron contrib-
Short IPI could also be a marker for cultural, family utes to neurotransmitter production, myelination, and
planning or lifestyle factors [Cheslack-Postava et al., immune function [Beard, 2000], processes shown to be
2011; Dodds et al., 2011], and evidence for a U-shaped dysregulated in autism [Beard, 2000].
relationship between IPI and ASD supports a contribu- Results from two large studies of maternal iron during
tion from such influences [Cheslack-Postava et al., pregnancy and ASD risk are inconsistent. In the
2014; Durkin et al., 2015; Zerbo et al., 2015]. California-based CHARGE case-control study (520 cases,
346 typically developing controls), children of mothers
with the highest versus lowest iron intake from supple-
Iron. Iron is critical for fetal and placental growth ments and cereals had half the odds of developing ASD,
and brain development and functioning. During preg- especially when mothers were of advanced age, obese,
nancy, iron deficiency, which is common [O’Brien, diabetic, or hypertensive [Schmidt, Tancredi, Krako-
Zavaleta, Abrams, & Caulfield, 2003; Stoltzfus, 2001] wiak, Hansen, & Ozonoff, 2014b]. In contrast, in the
can induce fetal and infant iron-deficiency [Allen, 2000; Norwegian birth cohort study [Suren et al., 2013], ASD
Colomer et al., 1990; Millard, Frazer, Wilkins, & Ander- risk did not differ for those taking vitamin and mineral
son, 2004; Tchernia, Archambeaud, Yvart, & Diallo, supplements likely to contain iron, compared with
1996]. Low iron has been associated with developmen- those taking no vitamins or minerals. The studies dif-
tal delays and behavioral disturbances including fered on obesity prevalence; potential for recall bias;

10 Hertz-Picciotto et al./Environmental contributions to ASD INSAR


Licensed to Mariana Wollinger Maciel Fassheber - marianawmaciel@gmail.com - 046.925.309-60

and measure of supplemental iron intake. Neither study Syndrome Test examined fish intake collected prospec-
accounted for total dietary iron intake or maternal iron tively; the researchers reported consumption of fatty
status, concerns that future studies should heed. fish above the recommended limit during pregnancy
(>340g/week) was associated with fewer autistic traits
along with higher cognitive scores [Julvez et al., 2016].
Fatty acids and fish. Polyunsaturated fatty acids
This was the first study to control for confounding by
(PUFAs), especially omega-3, play significant roles in
mercury (measured in umbilical cord blood). Additional
the structure, function, and development of human
strengths were the large sample, inclusion of both die-
brains [Freeman et al., 2006; McNamara & Carlson,
tary and supplement intake, and analyses by timing,
2006; Rombaldi Bernardi, de Souza Escobar, Ferreira, &
which showed stronger associations for prenatal as
Pelufo Silveira, 2012]. Differential PUFA status in indi-
compared with childhood intake. Conclusions on fish
viduals with versus without ASD [Brigandi et al., 2015;
and PUFAs will require more studies that include sup-
Jory, 2016; Wiest, German, Harvey, Watkins, & Hertz-
plements and diet, adjust for fish contaminants, and
Picciotto, 2009; Yui, Imataka, Kawasaki, & Yamada,
consider exposure windows.
2016], and small studies of fatty acid supplementation
to improve ASD symptoms were recently reviewed
[Mazahery et al., 2017]. Omega-3 consumption during Smoking. The literature on maternal smoking and
pregnancy has been associated with higher IQ [Richard- ASD has been contradictory, and two meta-analyses
son, Easton, & Puri, 2000b] and fewer social problems that employed random effects models on fifteen studies
[Richardson & Ross, 2000]. Similarly, lower omega- reported the same estimated OR 5 1.02 and almost
3:omega-6 ratio in maternal blood was associated with identical CI (0.93, 1.12) [Rosen, Lee, Lee, Yang, & Bur-
poor communicative development [Strain et al., 2008] styn, 2015] and (0.93, 1.13) [Tang, Wang, Gong, &
Maternal seafood and fish intake, the primary source of Wang, 2015]. Sensitivity analyses from both teams
PUFAs [Peet, Laugharne, Mellor, & Ramchand, 1996; found the results to be robust with no evidence of pub-
Richardson et al., 2000a], improves child neurodevelop- lication bias. Is this sufficient evidence of no effect to
mental outcomes (reviewed in [Avella-Garcia & Julvez be definitive? Two concerns detract from that conclu-
2014; Starling, Charlton, McMahon, & Lucas, 2015]), sion. First, of the 15 studies included in the analysis by
however, positive impacts may be attenuated by con- Tang et al. [2015], six did not adjust for any confound-
taminants such as mercury and PCBs, which bioaccu- ers, and four adjusted for either birthweight or gesta-
mulate in these foods [Avella-Garcia & Julvez, 2014]. tional age, both of which are strongly associated with
The few studies that examined maternal fatty acids smoking and ASD and could be intermediates, for
and ASD risk are not concordant. The Norwegian study which a standard adjustment as a confounder can intro-
found no association of prenatal fish oil supplements duce bias. Second, virtually all studies ascertained
with ASD risk [Suren et al., 2013]. In contrast, mothers smoking by self-report, whether to a mid-wife, on a
in the Nurses’ Health Study II who reported having a standardized questionnaire, or for a birth certificate;
child with autism also reported diets lower in omega-6, birth certificates have long been known to provide
linoleic acid, and total PUFAs, and very low intake (the underestimates of smoking prevalence, and more gener-
lowest 5%) of omega-3 [Lyall, Munger, O’Reilly, Santan- ally, self-report of smoking during pregnancy is subject
gelo, & Ascherio, 2013], but similar intake of fish to to significant under-reporting. Gold-standard biochemi-
mothers not reporting a child with ASD. However, die- cal measures of active smoking indicate about 25%
tary information was not collected for the pregnancy underascertainment using self-report. [Dietz et al.,
period for most participants, PUFA from supplements 2011; Kang et al., 2013; Shipton et al., 2009; Swamy,
was not available, and contaminants in fish were not Reddick, Brouwer, Pollak, & Myers, 2011]. This high
considered. level of misclassification could bias results towards a
Two studies assessed autistic traits. The Generation R no-effect finding.
Cohort of the Netherlands obtained the Social Respon-
Medical Factors (Table 3)
siveness Scale on >4,600 children, and measured PUFAs
at midpregnancy: lower maternal plasma omega- Medications. Described above, prenatal valproic acid
3:omega-6 ratio, and higher total omega-6 and linoleic and thalidomide exposures predicted strongly elevated
acid were associated with more autistic traits at 6 years risk for autism. More recently, some reports have sug-
[Steenweg-de Graaff, Ghassabian, Jaddoe, Tiemeier, & gested that use of SSRI’s may, in certain subsets, be
Roza, 2016]. Maternal omega-3 status and prenatal die- associated with ASD [Croen, Grether, Yoshida, Odouli,
tary fish intake were not associated with autistic traits. & Hendrick, 2011; El Marroun et al., 2014; Harrington
This study did not account for mercury or PCB expo- et al., 2014; Rai et al., 2013], with two studies suggesting
sure. A study that administered the Childhood Asperger greatest risk with first trimester exposures [Croen et al.,

INSAR Hertz-Picciotto et al./Environmental contributions to ASD 11


Licensed to Mariana Wollinger Maciel Fassheber - marianawmaciel@gmail.com - 046.925.309-60

Table 3. Evidence on Medical Conditions, Procedures, or Medications in Relation to Autism Spectrum Disorder or Autism
Symptoms
Medical Factors Summary of evidence Influential References or Reviews

Thalidomide Case reports of increased autism prevalence in children of mothers given Stromland et al. [1994]
Valproic acid/anti- these drugs during pregnancy. With low prevalence of use, these factors Moore et al. [2000]; Rodier et al. [1996]
epileptic drugs can account for very few cases of autism today.
Selective Serotonin Most studies found an association of SSRI use during pregnancy and Croen et al. [2011]; Hviid et al. [2013]; El
Reuptake Inhibitors increased risk in either a specific time period or a specific subset. Meta- Marroun et al. [2014]; Harrington et al.
analysis of four case-control studies produced and adjusted OR51.8 (95% [2014]; Rai et al. [2013]; Malm et al.
CI 5 1.5, 2.2). As with individual studies, confounding by indication could [2016]; Man et al. [2015]
not be excluded. Studies attempting to assess such confounding yielded
conflicting results.
Pre-eclampsia Several large studies provide evidence of increased risk for ASD. Studies Walker et al. [2015]; Mann et al. [2010];
relied on (a) administrative medical databases (e.g., Medicaid), or (b) Burstyn et al. [2010]; Buchmayer et al.
detailed abstraction of medical records including placental insufficiency. [2009]; Dodds et al. [2011]
The latter produced a larger effect size than administrative databases,
possibly because detailed abstraction of notes in medical charts beyond
standard codes, is more complete. RR’s ranged from 1.2 to over 2.
Gestational diabetes High degree of consistency in associations with ASD across studies, includ- Krakowiak et al. [2012]; Lyall et al. [2012];
ing two meta-analyses, several large cohorts, and a few moderate-sized Xiang et al. [2015]; Connolly et al. [2016];
studies. Meta-analysis, two large cohorts and one moderate-sized case- Li et al. [2016]; Nahum Sacks et al. [2016]
control study found RRs/OR of 1.3–1.5; two studies with <30 cases Meta-analysis: Xu et al. [2014]
observed RRs of 3–4 (with wide CI’s). Strong biologic plausibility; co-
occurring obesity may exacerbate effect. Whether controlled versus uncon-
trolled diabetes alters association has not been examined.
Maternal infections/fever Congenital rubella, maternal fever and influenza, and other viruses during Chess et al. [1977]; Desmond et al. [1967];
during pregnancy pregnancy are each associated with autism in multiple studies and in Deykin and MacMahon [1979]; Zerbo
meta-analysis. In two studies, the impactful timing of maternal fever was et al. [2012]; Atladottir [2010, 2012];
2nd trimester, and limited to fevers not treated with antipyretics. Addi- Webb et al. [2017]; Hornig et al. [2017]
tionally, copy number variants and viral infection combined appear to
increase severity of autism symptoms.

2011; Harrington et al., 2014]. A meta-analysis of nine no estimate of duration or intensity of the thermal
studies produced pooled ORs for case-control studies and index, suggest that the exposure metric may have done a
cohort studies, respectively, of 2.1 (95% CI, 1.7, 2.7) and poor job of rank ordering children by their true doses. A
1.8 (1.5, 2.2). As with many of the individual studies, second null study randomized pregnancies to different
this analysis was subject to confounding by indication. ultrasound and Doppler imaging protocols, but took
However, larger studies that did take such confounding place before introduction of higher dose equipment
into account by comparing depressed mothers who did [Stoch et al., 2012]. The third investigation evaluated the
vs. did not medicate (or did not medicate with SSRIs) triple-hit hypothesis, which proposed ASD phenotypes
produced conflicting results [Malm et al., 2016; Rai et al., to be dependent on three insults combined: (a) environ-
2013]; neither study examined timing of exposure. Nev- mental exposures; (b) their specific timing; and (c)
ertheless, mother’s mental state can alter the intra- genetic susceptibility [Williams & Casanova, 2010]. This
uterine (and postnatal) environment for the child. At concept was operationalized for exposures to 1st trimes-
this stage, no conclusion can yet be drawn as to whether ter ultrasound and presence of CNV in an ASD popula-
SSRIs influence risk of ASD. tion (n 5 1749) from the Simons Simplex Collection
[Webb et al., 2017]. Those with vs. without a 1st trimes-
ter ultrasound scored significantly lower in social affect,
Medical interventions. Use of ultrasonography for and significantly higher in repetitive/restricted behav-
imaging the fetus is now nearly universal. Both fre- iors. Limiting to males with CNV, the 1st trimester ultra-
quency of use and thermal intensity have increased, the sound associated with significantly lower nonverbal IQs,
latter by 8-fold during the 1990s as a result of FDA lifting and higher repetitive behavior scores. Without a control
prior restrictions [Webb et al., 2017]. A few studies group, ASD risk could not be assessed, but results suggest
addressed ultrasound in relation to ASD. One found no 1st trimester ultrasound may influence the severity of
association in analyses of the number of ultrasounds impairments, particularly in those with underlying
received, but the study period for these exposures (1994– genetic susceptibility.
1999) covered the changes in equipment [Grether, Li, Infertility treatments have also received attention.
Yoshida, & Croen, 2010]. That problem, combined with Because both ASD and each type of treatment are both

12 Hertz-Picciotto et al./Environmental contributions to ASD INSAR


Licensed to Mariana Wollinger Maciel Fassheber - marianawmaciel@gmail.com - 046.925.309-60

rare, few if any studies have been sufficiently powered Much work has examined maternal diabetes during
[Lyall et al., 2017a]. pregnancy as a risk factor for ASD with or without obe-
sity [Connolly et al., 2016; Krakowiak et al., 2012; Li
et al., 2016; Xiang et al., 2015], explored in greater
Acute illnesses. Decades ago, two independent stud-
depth below.
ies demonstrated a strong association of ASD with con-
genital rubella [Chess, 1977; Desmond et al., 1967]. In
another early epidemiologic study, prenatal influenza
Methodologic Challenges
or other viral infection appeared to predispose the off-
spring to develop autism [Deykin & MacMahon, 1979]
To highlight methodological challenges and illustrate
and maternal influenza was recently implicated in an
conceptual issues surrounding environmental epidemio-
investigation from Denmark [Atladottir, Henriksen,
logic studies, three investigations from the CHARGE
Schendel, & Parner, 2012], with a doubling of risk. In
Study are described. One of the linchpins of this field is
another report, these researchers described 3-fold
the difficulty capturing exposures accurately, which is
increased autism risks after viral infection in the 1st tri-
especially acute for case-control studies. For the exam-
mester, or 40% higher bacterial infection in the 2nd tri-
ple of pesticides, we demonstrate one approach. A sec-
mester [Atladottir et al., 2010]. Two studies, including
ond challenge is that of causal inference in the context
one prospective cohort, observed an association of ASD
with fever in the 2nd trimester, primarily when not of risk factors for which a randomized controlled clini-
treated with an anti-pyretic [Hornig et al., 2017; Zerbo cal trial is not an option for either ethical reasons or
et al., 2012]. Prolonged fever (a week or longer) feasibility. The field of epidemiology has developed sev-
appeared to confer a 3-fold higher risk of autism in the eral frameworks to ensure that inferences are based on
Danish study. Supporting a role for acute infectious ill- sound science, logical reasoning and transparency; we
nesses or fever in ASD is a growing literature on inflam- illustrate with the case of maternal diabetes. From the
mation or immune dysregulation in relation to autism third example, maternal folic acid intake, we explore
[Braunschweig et al., 2012, 2013; Gesundheit et al., issues of gene-environment interaction and epigenetics.
2013; Hsiao, 2013; Vargas, Nascimbene, Krishnan, Zim- Briefly, the CHARGE Study was funded in 2002 to
merman, & Pardo, 2005] and on animal models involv- identify causes and contributing factors for ASD, along
ing infection or immunological perturbations similar to with other developmental delays, with a focus on envi-
those induced by infection and resulting neurobehavio- ronmental chemicals and their interactions with
ral aberrations (see review emphasizing timing [Meyer, genetic susceptibility. The case-control design is well-
Yee, & Feldon, 2007; Patterson, 2011]. Studies of sea- suited for outcomes with poorly understood etiology, as
sonality of birth or conception add circumstantial evi- this approach allows for multiple exposures to be inves-
dence insofar as influenza and other infectious diseases tigated, and is more efficient for rare conditions (e.g.,
also follow seasonal patterns. Norwegian mothers of <5% risk) and those with a potential latency (time
male children with ASD versus male controls were between exposure and diagnosis) of years, than a pro-
observed to have higher mid-pregnancy serum anti- spective cohort. With prenatal exposures of special con-
bodies to Herpes simplex virus-2 [Mahic et al., 2017], cern and diagnoses frequently made at ages 3 or
however, the proportion who were seropositive did not beyond, the case-control design also promised statisti-
differ and several potential confounders were not con- cally powerful results within a shorter time frame.
trolled. A previous study found no association with Three groups of children were enrolled (ASD, other
HSV-2 [Atladottir et al., 2012]. developmental delays without ASD symptoms (DD),
and controls from the general population), and final
Chronic conditions, inflammation. A literature on diagnoses of ASD, DD or typical development were
complications of pregnancy and delivery has emerged. based on a full clinical assessment (for further protocol
Hypertensive disorders of pregnancy including pre- details, see [Hertz-Picciotto et al., 2006]). To address the
eclampsia have been associated with ASD, consistently primary limitation of case-control studies—retrospective
in large, well-conducted studies [Buchmayer et al., assessment of exposure—CHARGE obtained, whenever
2009; Burstyn, Sithole, & Zwaigenbaum, 2010; Mann, feasible, objective measures. Two sources are described
McDermott, Bao, Hardin, & Gregg, 2010; Walker et al., in the examples below: a database of environmental
2015]. Preeclampsia explained much of the association exposures, and medical records. Both carry information
between preterm delivery and ASD [Buchmayer et al., collected in real-time for specific time windows. Such
2009] and remained strongly associated even after historical data constitute strengths, essentially obviat-
adjustment for intermediates, gestational age, or birth- ing this limitation of the case-control design. With
weight [Mann et al., 2010]. these types of exposure data, a case-control study can

INSAR Hertz-Picciotto et al./Environmental contributions to ASD 13


Licensed to Mariana Wollinger Maciel Fassheber - marianawmaciel@gmail.com - 046.925.309-60

Table 4. Prospective Studies of Prenatal Measures of Pesticide Exposures and Child Neurodevelopmental Outcomes Other
than ASD
Outcome Study Population Reference

Memory and IQ New York inner-city children age 3 and age 7 years Rauh et al. [2006, 2011]
Brain volumetric measurements Same cohort, age 8–9 years PNAS, Rauh et al. [2012]
Tremor Same cohort, age 11 years Neurotoxicology, Rauh et al. [2015]
Mental & psychomotor development California farmworker children, age 24 months Eskenazi et al. [2007]
Memory and IQ Same cohort, age 7 years Bouchard et al. [2011]
Attention/ADHD Children ages 36 months and 5 years Marks et al. [2010]
Mental and psychomotor development and reasoning Urban child population, ages 12 months and 6–9 years Engel et al. [2011]

provide exposure assignments equal in quality to those doubled with a ten-fold increase in metabolites [Eske-
obtained in prospective studies. nazi et al., 2007].
Two case-control studies examined ASD in association
Example #1-Pesticides: The challenge of exposure assessment
with distance of home to agricultural pesticide applica-
Humans exposures to pesticides occur from: (a) insect tions. All commercial pesticide applications in Califor-
or other pest control products (sprays, bombs, etc.) in nia agriculture are reportable by law to the Department
homes, malls, restaurants, or on pets and gardens, even of Pesticide Regulation, including specific chemicals
when used according to instructions; (b) drift from used, amounts, dates of applications, and locations.
applications on orchards or agricultural fields, well- Both studies used Geographic Information Systems to
documented at distances of several kilometers from the overlay maternal addresses with locations of the pesti-
application site [Wofford et al., 2014]; and (c) residues cide applications from this database. Amounts of pesti-
in fruits, vegetables, and, through bioaccumulation, cide applied within pre-defined distances of the
meats, fish, and dairy products [Schafer & Kegley, 2002; residences were compared for cases vs. controls. One
Vogt et al., 2012]. study showed increased risks for residential proximity
Most pesticides are designed to be neurotoxic. Orga- to two organochlorine pesticides in the first trimester,
nophosphates, currently among the most abundantly and for applications of organophosphates and a specific
used insecticides, are acutely toxic due to inhibition of pyrethroid (bifenthrin) at any time during pregnancy
the enzyme acetylcholinesterase (AChE), which breaks [Roberts et al., 2007].
down neurotransmitters at the post-synaptic mem- The second case-control study used CHARGE partici-
brane. However, neurodevelopmental and neurologic pants [Shelton et al., 2014], for whom all diagnoses
dysfunction occur at subacute levels, when AChE inhi- were clinically confirmed. The three chemical classes
bition is absent, [Voorhees, Rohlman, Lein, & Pieper, investigated were: organophosphate, pyrethroid, and
2016]. Hence other toxic mechanisms are at play and carbamate pesticides. Figure 3 demonstrates linkage of
potentially relevant for gestational exposures in early pesticide use reports to the residences. All addresses
brain development, including inflammation, GABA sig- from 3 months before conception through pregnancy
naling, thyroid disruption, oxidative stress and mito- were geocoded and buffers of radii 1.25, 1.5, or 1.75
chondrial dysregulation [Banks & Lein, 2012; Schuh, kilometers were created. Overlaying on this map were
Lein, Beckles, & Jett, 2002; Shelton, Hertz-Picciotto, & the locations of pesticide applications, with each child
Pessah, 2012; Voorhees et al., 2016]. Prenatal organo- assigned exposures within the buffer zone in the win-
phosphate exposures have been associated with cogni- dows of interest: the 3 months before conception, each
tive impairments, attention deficits, and tremor in trimester, or the entire pregnancy. One salient finding
children (Table 4) [Bouchard et al., 2011; Engel et al., was the significant association between organophos-
2011; Rauh et al., 2011]. Some deficits persist into phate exposures at any point during pregnancy and
school ages. 60% higher risk for ASD (Fig. 4, top two panels). The
Three studies evaluated prenatal organophosphate strongest association was for 2nd and 3rd trimester
exposures and either an ASD diagnosis or symptoms. chlorpyrifos (an organophosphate), with a 2.5- to 3-fold
Urine specimens collected from pregnant women in a elevation in ASD risk. Chlorpyrifos was banned in 2001
farmworker community were analyzed for organophos- by the U.S. EPA for household products, and the safety
phate pesticide metabolites. When the children were 24 of its continued use in agriculture has been contested
months of age, pervasive developmental delay (PDD) based on both toxicologic and epidemiologic research.
symptoms were assessed. The proportion scoring In fall 2016, U.S. EPA determined that residues in food
>97.5th percentile was unexpectedly high (possibly and water exceeded standards for safe levels. In March
related to administration of the scale in Spanish) and 2017, the U.S. EPA decided not to follow through on its

14 Hertz-Picciotto et al./Environmental contributions to ASD INSAR


Licensed to Mariana Wollinger Maciel Fassheber - marianawmaciel@gmail.com - 046.925.309-60

of the pesticides in air samples is sufficient to presume


human exposures, since even indoor air reflects outdoor
air pollution and only those who leave the area for
extended time periods would avoid breathing the air.
Although for some chemicals biologic measures may be
preferable, the organophosphate and pyrethroids com-
pounds are quickly metabolized and excreted. Hence,
even multiple specimens can easily fail to capture an
exposure that may have damaged the developing brain.
Residential proximity to applications recorded in real-
time not only is reasonable and now established as valid,
but may be an optimal exposure metric as it accounts for
exposures throughout preconception and pregnancy.
Further support for long-lasting effects from prenatal
pesticide exposures is a small MRI study in the Colum-
bia Children’s Cohort. Chlorpyrifos was measured in
3rd trimester blood samples. Twenty children, whose
mothers were nonsmokers with low exposure to air pol-
lution, underwent MRI at ages 8–9 years. Among those
Figure 3. Household linkage to Pesticide Use Report (PUR, a with high (vs. low) chlorpyrifos, cortical regions of the
database of pesticide applications based on mandated reporting brain responsible for attention, receptive language proc-
by commercial applicators). The dot represents a residence,
essing, social cognition and inhibitory control were
each square on the grid is 1 3 1 mile, representing the
township-range-section for which pesticide application loca- enlarged [Rauh et al., 2012]. These represent core ASD
tions are geocoded, and the circle is a buffer around the house, symptoms and common comorbid deficits, adding bio-
used for assigning exposures. Shelton et al. [2014], online sup- logic plausibility to the pesticide-ASD link.
plementary material. (This figure was originally posted in the Thus, two studies used an objective validated measure
online supplementary material of this Open Access journal.). of exposure with accurate timing. A third obtained a
biomarker prospectively. In total, three studies impli-
own proposed rule to ban use of chlorpyrifos in agricul- cate prenatal pesticides in ASD: each had a sizable sam-
ture [U.S. Environmental Protection Agency, 2017]. ple, adjusted for confounding, and used an objective
Additionally, preconception pyrethroid exposures measure of exposure specific to a period in pregnancy.
(Fig. 4, lower two panels) showed an 80% increased risk The two that examined pyrethroids both found associa-
for ASD. Marketed as a natural product (pyrethrins are tions, and all three observed associations of ASD with
produced by the chrysanthemum plant), pyrethroids are organophosphate pesticides or their metabolites.
synthetic compounds chemically engineered to be
more toxic and to last longer. Pyrethrins rapidly Example #2-Maternal diabetes: Causal inference in
degrade in sunlight whereas bifenthrin has a half-life in non-randomized studies
the outdoors of about a year.
In environmental health studies, assessment of expo- Obesity prevalence has increased at all ages, and by
sure stands out as the critical component. Objections 2010, over one-third of U.S. adults were obese [Flegal,
that pesticide applications do not reflect exposure [Burns, Carroll, Kit, & Ogden, 2012]. Over 20% aged 20–39
Cohen, & Lunchick, 2015] were not supported by a have metabolic syndrome [Ervin, 2009; Flegal, Carroll,
recent year-long environmental monitoring study con- Ogden, & Curtin, 2010], diabetes prevalence doubled
ducted in the California central valley. Air measurements from 1980 to 2012 [Geiss et al., 2014], and among preg-
of chlorpyrifos in a town located near agricultural fields nant women in California, 9% were diagnosed with
exhibited a direct variation with the amount of this pesti- type 2 or gestational diabetes [Lawrence, Contreras,
cide applied within an 8 km buffer (Fig. 5) [Wofford et al., Chen, & Sacks, 2008], while 5% of U.S. pregnancies in
2014]. Similar high correlations were observed for air 2009 were complicated by hypertensive disorders [Mar-
concentrations of numerous other pesticides compared tin et al., 2011]. Against this backdrop, a growing num-
with amounts known to have been applied, establishing ber of studies has observed an association between
drift at distances similar to those used as buffers in maternal prenatal diabetes and ASD [Connolly et al.,
research studies on child development. Wofford et al. 2016; Krakowiak et al., 2012; Li et al., 2016; Nahum
thus demonstrate that pesticide applications serve as a Sacks et al., 2016; Xiang et al., 2015; Xu, Jing, Bowers,
valid measure of exposures to nearby residents. Presence Liu, & Bao, 2014].

INSAR Hertz-Picciotto et al./Environmental contributions to ASD 15


Licensed to Mariana Wollinger Maciel Fassheber - marianawmaciel@gmail.com - 046.925.309-60

Figure 4. Associations of agricultural pesticide applications with risk of autism spectrum disorder (ASD). Graphs present the odds
ratios (OR’s) comparing ASD cases to typically developing controls with respect to their exposures to organophosphate (upper two
panels) and pyrethroid (lower two panels) pesticides. Each column represents a different time window and results for the three dif-
ferent buffers (1.25, 1.5, and 1.75 km) are within each box, with a solid square representing the OR and a bar for the 95% confi-
dence interval. OR’s are plotted on the logarithmic scale.

whose mothers’ pregnancies were complicated by one


or more metabolic conditions were 60% more likely to
have ASD and over twice as likely to have DD compared
with children whose mothers had healthy weight and
no metabolic conditions. These results were adjusted
for sociodemographic characteristics, calendar time,
and the child’s age and sex [Krakowiak et al., 2012].
Obesity, the most prevalent metabolic condition, drove
these associations. Metabolic conditions involve dysre-
gulation marked by elevated glucose, triglycerides, cho-
lesterol, leptin, and proinflammatory immune markers.
Maternal diabetes, primarily gestational, was associ-
Figure 5. Comparison of average weekly concentrations of ated with a doubled risk for DD and a 50% increased
chlorpyrifos detected at three monitoring sites in a rural Califor- likelihood of ASD although the latter did not reach sta-
nia town over a one-year period, and total reported use of tistical significance. An earlier meta-analysis had sug-
chlorpyrifos by week in an 8 km study area surrounding the gested an ASD-maternal diabetes association [Gardener,
monitoring sites. Reproduced from Wofford et al. [2014]. Spiegelman, & Buka, 2009], and recently, a meta-
(Reprinted with permission from Springer.). analysis of 12 studies (9 case-control, including the
CHARGE Study; 3 cohort) also demonstrated signifi-
In the CHARGE Study, compared with mothers of cantly elevated risks (cohort: pooled RR 5 1.5, 95% CI
typically developing children, those of children with 1.3, 1.8; case-control: pooled OR 5 1.7, 95% CI 1.2, 2.4)
ASD or other DD were disproportionately afflicted with [Xu et al., 2014]. Newer large epidemiologic studies also
metabolic conditions during pregnancy, including obe- confirm the association between maternal diabetes
sity, diabetes (type 2, gestational), and underlying (alone or in combination with obesity) and risk for ASD
hypertension or preeclampsia [Krakowiak et al., 2012; [Connolly et al., 2016; Li et al., 2016; Xiang et al.,
Walker et al., 2015]. Among nearly 1,000 children clini- 2015]. In a retrospective cohort of 322,000 births at an
cally assessed using standardized instruments, those HMO [Xiang et al., 2015], both pregestational and

16 Hertz-Picciotto et al./Environmental contributions to ASD INSAR


Licensed to Mariana Wollinger Maciel Fassheber - marianawmaciel@gmail.com - 046.925.309-60

which maintain a proinflammatory environment and


reduce insulin sensitivity. If compensatory mechanisms
fail, maternal hyperglycemia may ensue resulting in an
oxygen-deprived fetal brain at a critical time [Burstyn,
Wang, Yasui, Sithole, & Zwaigenbaum, 2011; Eidelman
& Samueloff, 2002]. Other pathways may involve meta-
bolic and immune derangements, stimulating placental
proinflammatory responses and inducing oxidative
stress. Oxidative stress has been linked to ASD [Chau-
Figure 6. Causal model linking maternal metabolic conditions han, Chauhan, Brown, & Cohen, 2004; Yao, Walsh,
to environmental exposures, biomarkers in the newborn, bio- McGinnis, & Pratico, 2006], and skewed immune
markers in childhood, gene expression, and diagnosis. Bold marker profiles have been described in newborns and
arrows represent associations that have been published [Krako- children with ASD [Krakowiak et al., 2017a; Onore,
wiak et al., 2012; Krakowiak et al., 2017a; Krakowiak et al., Careaga, & Ashwood, 2012]. Additionally, mothers may
2017b]. experience a breach in immune tolerance from meta-
gestational diabetes diagnosed at 26 weeks gestation bolic dysfunction. Approximately 23% of mothers of
were significantly associated with 33% to 42% greater children with ASD produce anti-fetal brain autoantibod-
ASD risk. Major strengths of the study were: adjustment ies compared with 1% of mothers of typically develop-
for critical confounders; medical records and glucose ing children [Braunschweig et al., 2012, 2013]. Among
values to confirm diabetes diagnoses; child follow-up mothers whose child had ASD, we found gestational
for a median 5.5 years; and diagnoses by pediatric diabetes to be associated with a 3-fold increased preva-
developmental specialists. Results were similar in a large lence of having these autoantibodies [Krakowiak,
cohort of over 40,000 children [Connolly et al., 2016], Walker, Tancredi, Hertz-Picciotto, & Van de Water,
which showed diabetes and obesity associated with, 2017b]. Both ASD and maternal anti-fetal brain autoan-
respectively, 56% and 50% greater risks for ASD; more- tibodies have been independently associated with the c-
over, offspring from pregnancies complicated by both MET “C” allele polymorphism, which regulates immune
gestational diabetes and obesity had a 2.5-fold increased response [Campbell et al., 2008; Heuer et al., 2011;
likelihood of ASD. A large sample size and meticulous Jackson et al., 2009]. In animal models, impaired MET
protocol for case ascertainment from EMR were major signaling is associated with diabetes [Araujo et al.,
strengths of this study. The reliance on birth certificates 2012; Demirci et al., 2012]. Additionally, many ASD
for maternal diabetes and height and weight (to calcu- genes involve insulin signaling pathways (e.g., PI3K/Tor
late obesity) is not optimal, though prevalences of these (PI3K and mTOR)) [Stern, 2011]. Finally, the maternal
conditions were comparable to other studies, suggesting diabetic environment could alter the placental epige-
underascertainment was not major. In a smaller low- nome and ultimately its regulation of fetal develop-
income urban cohort from Boston, children of mothers ment [Marsit, 2016]. Overall, numerous mechanisms
with gestational or pregestational diabetes accompanied can plausibly link maternal diabetes with disrupted
by obesity had a 3 to 4-fold increased risk for ASD,
brain development; moreover, biomarkers for these per-
adjusting for confounders [Li et al., 2016]. The
turbations have been found, in some cases prior to the
strengths were 6 years follow-up and medical records of
ASD diagnosis.
both maternal and child diagnoses. A fourth study
involved a cohort of over 230,000 births in which out-
comes were hospitalizations with one or more neuro- Inferring causation when randomized trials are
psychiatric diagnosis, of which ASD was one [Nahum inappropriate, unethical, or unfeasible. Conducting
Sacks et al., 2016]. A 4-fold increased risk was associated randomized trials for studying environmental risk fac-
with maternal gestational diabetes, but the number of tors is often impossible. An intervention that increases
hospitalized cases of ASD was small. exposures suspected to cause harm would be unethical.
When exposures are present in diverse sources, not
Underlying biological mechanisms. Several mecha- revealed in product labels (e.g., cosmetics, air fresh-
nisms have been hypothesized to explain the link of eners, etc.), and/or have outcomes that take years to
maternal diabetes with offspring ASD. Metabolic condi- manifest, even an intervention to reduce exposures
tions induce persistent low-grade inflammation and may be infeasible. For these reasons, epidemiologists
insulin resistance [Makki, Froguel, & Wolowczuk, 2013; rely on observational studies, which makes inference of
Pantham, Aye, & Powell, 2015]. Such inflammation causation from challenging, though not impossible.
leads to abundant immune cells in adipose tissue, Indeed, many etiologic factors have been established

INSAR Hertz-Picciotto et al./Environmental contributions to ASD 17


Licensed to Mariana Wollinger Maciel Fassheber - marianawmaciel@gmail.com - 046.925.309-60

from nonrandomized investigations (e.g., cigarette and (g) genetic pathways related to both ASD and dia-
smoking and lung cancer). betes or insulin regulation offer further evidence for
When does evidence from observational associations coherence with known facts. The literature provides
permit causal inference? Since the purpose of randomi- substantial support for prenatal maternal diabetes as a
zation is to control for systematic differences (con- causative factor in ASD, based on the Hill viewpoints.
founders) between treated and untreated groups,
Example #3-Folic Acid: Integration of environment, genetics
identification of confounders and their control is at the
& epigenetics
core of the epidemiologic approach. When epidemio-
logic studies are referred to as ‘correlational’, this is usu- Periconceptional folic acid (FA) supplements can pre-
ally an inaccurate characterization since a correlational vent up to 70% of neural tube defects (NTDs). From
study by definition does not take into account con- large scale RCTs [MRC Vitamin Study Research Group,
founders, and hence would fail the most elementary 1991], first trimester folic acid supplementation is rec-
test of rigor. More generally, methodologic soundness ommended worldwide to prevent neural tube defects
of individual studies depends on: reliable, valid expo- (NTDs): 4–5 (0.4–1) mg/day for women with (without)
sure measures, high quality diagnostic or dimensional a history of NTDs [Centers for Disease Control, 1991].
outcome measures, a systematic approach to con- Cohort studies have also found that maternal FA sup-
founder control, minimization of selection bias, and plement intake before/during early pregnancy was asso-
statistical analyses appropriate to the nature of the ciated with fewer childhood behavioral problems [Roza
data. Additionally, the body of literature should be et al., 2010] or hyperactivity and peer problems [Schlotz
assessed in context of the Hill ‘viewpoints’ (first et al., 2010], improved verbal, verbal-executive func-
described for evaluating epidemiologic evidence that tion, attention, and social competence [Julvez et al.,
smoking caused lung cancer [Hill, 1965]), including:
2009], and decreased risk for severe language delays
temporality (the exposure precedes onset of the out-
[Roth et al., 2011].
come); strength of the association, consistency across
Previously, we found that maternal report of prenatal
studies in different populations, biologic plausibility,
vitamin intake near conception was associated with
and coherence with known scientific facts.
40% reduced risk of ASD in 707 California-born
Applying the above approach to the major studies
CHARGE Study children with autism or clinically-
described above on maternal diabetes during preg-
confirmed typical development. Given that mothers
nancy: (a) For exposure, all but one used medically
recalled their vitamin intake retrospectively, recall bias
reported diagnoses of diabetes, or self-report validated
could not be excluded. However, this result was repli-
by medical reports [Krakowiak, Walker, Tancredi, &
cated in a prospective birth cohort study with 85,176
Hertz-Picciotto, 2015], confirming both temporality and
Norwegian children [Suren et al., 2013]. Prenatal vita-
validity of the exposure; the one exception used birth
mins contain high levels of FA, which is converted to
certificates which appeared to show face validity based
folate under genetic regulation by MTHFR677A. Addi-
on the prevalences of these conditions [Connolly et al.,
2016]. (b) For outcomes: most used physician-diagnosed tional population-based cohort studies [Braun et al.,
cases from medical records, one used gold standard 2014a; Steenweg-de Graaff et al., 2015] also found pre-
research-reliable instruments, and one used neuropsy- natal vitamin or FA supplements associated with
chiatric hospitalization with ASD. Lacking a biologic reduced risk for autistic traits. However, when these
test for ASD, clinical diagnoses vary in how stringently studies examined maternal folate in serum collected after
DSM or ICD criteria are evaluated, which might lead to the first trimester, their results were null [Braun et al.,
some misclassification (in both directions), and, if non- 2014a; Steenweg-de Graaff et al., 2015], consistent with
differential relative to diabetes diagnosis, possibly may the literature on timing of folic acid protection against
explain substantially smaller associations in the larger ASD risk [Schmidt et al., 2011, 2012; Suren et al., 2013].
Kaiser and Ohio studies. (c) All studies used appropriate The only study not consistent with an early prenatal
multivariable models and (d) adjusted for maternal age, protective effect of FA was a large population-based
a candidate confounder; most also included year of birth cohort study in Denmark [Virk et al., 2016]. This
birth, gender, and race/ethnicity. (e) Although the study also failed to replicate one of the most established
strength of the associations varied, the two studies with risk factors for ASD, parental age, appearing to be an
large relative risks each had fewer than 30 exposed outlier.
cases [Li et al., 2016; Nahum Sacks et al., 2016], result- To date, only the CHARGE Study evaluated ASD and
ing in CI consistent with the smaller, more precise esti- FA intake by maternal and child genetics. We found
mates of the other studies; hence, along with earlier periconceptional maternal supplemental FA was associ-
meta-analyses, these results therefore demonstrate con- ated with reduced autism risk only when mothers and/
sistency. (f) Biologic plausibility was described above, or children had genotypes conferring inefficient folate

18 Hertz-Picciotto et al./Environmental contributions to ASD INSAR


Licensed to Mariana Wollinger Maciel Fassheber - marianawmaciel@gmail.com - 046.925.309-60

metabolism [Schmidt et al., 2012]. Thus, it was the arsenic in humans [Howe et al., 2014; Lambrou et al.,
combined impact of not taking a prenatal vitamin near 2012] and PM2.5 in a clinical trial [Zhong et al., 2017]
conception and the mother or child having gene var- have also implicated DNA methylation mechanisms.
iants producing less efficient folate-dependent one-car-
bon metabolism that put the child at particularly high
Future Directions
risk for developing autism [Schmidt et al., 2011].
The observed gene-environment interaction strength-
The first decade of concerted efforts to identify environ-
ens the evidence that the observed protection is not
mental/modifiable causes of autism has produced a
due to biased reporting of folic acid intake. First of all,
plethora of clues about risk and protective factors, with
because the finding was time-limited to the period
increasingly compelling evidence for a few: short inter-
around conception, mothers who had typical children
pregnancy intervals, maternal diabetes. The field has
would have had to systematically report an earlier con-
begun to mature, moving into its mid- to late child-
sumption of prenatal vitamins than when they actually
hood stage, where research is building and consolidat-
did initiate it, or mothers of affected children would
ing gains through replication of results and refinement
have had to systematically report a later than actual
of methods, even while still uncovering new connec-
time for initiating supplementation. However, this
tions. Also arising from this literature is the recognition
hypothesized biased reporting would have to occur only
of complexity in ASD etiology. Thus, we are poised to
among those genetically endowed with inefficient met-
approach adolescence, where new developments in
abolic profiles, an unlikely proposition.
Given the strength of the interaction between genes molecular biology, electronic connectedness and big
and nutrition, supplemental folic acid could help over- data create possibilities and opportunities to address
come ASD in susceptible subsets of the population; if some of the major perplexing challenges.
so, folic acid intake at levels higher than current recom- The Exposome
mendations could be useful for mothers with genetic or
other susceptibility factors, including environmental Taking lessons from genome-wide studies, environmen-
exposures, to reduce ASD risk. For example, our latest tal health scientists have begun to explore moving from
findings indicate risk for ASD from prenatal pesticide candidate exposures to more comprehensive analyses of
exposures was reduced in children whose mothers took the universe of human exposures. Although genetics is
folic acid in the periconception, as compared to those exceedingly complex, the DNA code is, for the most
whose mothers did not [Schmidt et al., 2017]. Thus, part, fixed for life. In contrast, environment has the
higher maternal folic acid intake in the periconception additional feature (curse, for researchers!) of being in
could not only compensate for genetic risk, but also constant change. Environmental health investigators
help protect against other environmental chemicals. are currently innovating ‘exposomic’ approaches.
The protective mechanisms of periconceptional FA One of the first environment-wide analyses assessed
for NTDs [Blom, Shaw, den Heijer, & Finnell, 2006], 266 unique environmental factors in relation to type 2
ASD [Schaevitz & Berger-Sweeney, 2012], and other diabetes defined by a fasting blood glucose test. Similar
developmental conditions remain unknown. Folate’s to GWAS, adjustment for multiplicity testing was
antioxidant properties [Joshi, Adhikari, Patro, Chatto- accomplished using false discovery rate, and results
padhyay, & Mukherjee, 2001], DNA methylating capac- were validated in four cohorts [Patel, Bhattacharya, &
ity [James et al., 2004, 2009] and role in DNA repair Butte, 2010]. Novel potential risk factors were discov-
[Duthie, 1999] are all candidates. Folate is a major ered and confirmed, including two organochlorines (a
methyl donor, suggesting epigenetic regulation near PCB and a pesticide), interesting in light of diabetes-
conception when the DNA methylome is largely ASD associations described above. For ASD research, the
stripped and then reestablished [Reik & Walter, 2001]. principal obstacle to applying similar statistical meth-
This timing is congruent with the protective associa- odologies is the lack of large amounts of environmental
tions observed in epidemiologic ASD studies [Blom, data collected during relevant time windows in robust
2009; Schmidt et al., 2011]. Alterations of maternal die- samples sizes.
tary methyl-donor concentrations during methylome New technologies in exposure sciences are providing
establishment near conception has long-term epigenetic solutions for such collection. Nontargeted chemical
and health outcomes implications [Cooney, Dave, & analyses, for example, via gas (or liquid) chromatogra-
Wolff, 2002; Heijmans et al., 2008; Steegers-Theunissen phy with quantitative time-of-flight mass spectrometry,
et al., 2009; Wolff, Kodell, Moore, & Cooney, 1998]. are identifying an increasingly large array of chemicals
Studies examining folic acid attenuation of effects of in various media: food packaging [Cherta et al., 2015],
environmental hazards on other outcomes, including river water for pharmaceuticals [Martinez Bueno, Ulas-
bisphenol A in mice [Dolinoy, Huang, & Jirtle, 2007], zewska, Gomez, Hernando, & Fernandez-Alba, 2012],

INSAR Hertz-Picciotto et al./Environmental contributions to ASD 19


Licensed to Mariana Wollinger Maciel Fassheber - marianawmaciel@gmail.com - 046.925.309-60

and organics in breast milk [Baduel, Mueller, Tsai, & 2015] and elastic net addresses correlated covariates,
Gomez Ramos, 2015]. Wearable sensors are being and allows the number of predictors to be larger than
deployed for a variety of applications, such as: detec- the number of observations. Network and pathway
tion of infant movements [Chen, Xue, Mei, Bambang models can characterize overlapping and/or inter-
Oetomo, & Chen, 2016; Zhu et al., 2015], physiologic changeable functionality, and also suggest potential
measurements in aging populations or athletes [Patel, synergistic effects produced by chemicals acting on dif-
Chen, & Butte, 2012], electrolyte and metabolite moni- ferent steps along the same or converging pathways. An
toring in sweat, tears and saliva [Bandodkar & Wang, NIEHS workshop recently emphasized the need for toxi-
2014] among others [Zheng et al., 2014]. Sensors for cologists, statisticians and epidemiologists to work col-
exogenous chemicals include silicone wristbands [Ham- laboratively, bringing their respective knowledge bases
mel, Hoffman, Webster, Anderson, & Stapleton, 2016], to the mixture problem [Taylor et al., 2016].
skin patches [Romanyuk et al., 2014], or hand-held
Gene-Environment Interactions
breathalyzers [Gouma et al., 2017] for children or adults
[Hammel et al., 2016]. Sensor technology offers high- Interestingly, although one-carbon metabolism genes
dimensional data collection that can dramatically that influence levels and activity of folate appear to act
advance the field of environmental epidemiology. synergistically with folic acid intake to increase ASD
Another component of the exposome is the micro- risk, these same genes have been missed in the vast
biome: the >90% of our cells that are not human, but number of autism GWAS reports, likely because their
symbiotically inhabit our bodies. Research on micro-
high-risk variants are too common. This raises a conun-
biome is revealing different intestinal microbiota in
drum—how can we systematically identify the joint
children with ASD versus those with typical develop-
contributions of genetic and environment to ASD? If
ment, such as high prevalence of Clostridium [De
the scenario seen for folic acid is not unique—and
Angelis, Francavilla, Piccolo, De Giacomo, & Gobbetti,
chances are that common genetic variants are largely
2015; Finegold et al., 2010; Parracho, Bingham, Gibson,
undiscovered—then potentially vast numbers of yet-
& McCartney, 2005]. These findings comport with
unidentified genes may increase ASD risk only in com-
greater prevalence of GI symptoms in ASD [Chaidez,
bination with exposures that have gone unmeasured in
Hansen, & Hertz-Picciotto, 2014]. Relevance for etiology
genetic studies. Exploration of genetic susceptibility in
remains unclear, although, as stated in our introduction
combination with environmental exposures promises to
to causal concepts, interventions that alter the trajec-
be fruitful, particularly for common variants.
tory after a diagnosis has been made may alleviate
To date, few studies have addressed gene-
symptoms and in some cases result in markedly
environment interaction in ASD by examining synergis-
improved behavior. Numerous research groups are seek-
tic combined effects for a specific environmental expo-
ing to develop probiotics to alter the species present
and their relative abundance in hopes these will ulti- sure (or a small set of exposures) and a genetic
mately help children with autism lead more productive susceptibility. Two used the CHARGE Study to examine
and socially connected lives. whether risks for ASD from nutrition or air pollution
As in the field of genomics, the twin problems of differed by genotypes with known biologic relevance to
nonindependence (of genes, or of environmental fac- the selected exposure, comparing children with ASD to
tors) and mixtures present major statistical challenges. typically developing controls [Schmidt et al., 2011,
Models that assume independence continue to be used, 2012; Volk et al., 2014]. Two others used the Simons
despite evidence to the contrary (for both genes and Simplex cohort of affected individuals to investigate
environmental chemicals). Hierarchical regression and associations of a medical factor, in the presence of
Bayesian methods have addressed nonindependence CNVs, with severity scores [Mazina et al., 2015; Webb
across a large set of explanatory variables [Witte, Green- et al., 2017]. All four of these studies found evidence
land, Haile, & Bird, 1994], machine learning algorithms supporting the hypothesized interactions, and three of
accounting for nonindependence are being developed the four either focused on a pre-specified time period or
[Zou & Hastie, 2005], as are corrections for nonindepen- examined several in which associations were stronger
dence of alleles [Li et al., 2012]. For real-world scenarios for exposures in specific windows.
of simultaneous exposures, various tools address differ- Big-data approaches can harness genome-wide data in
ent questions: principal components can identify clus- combination with environmental factors for discovery-
ters of correlated items, reducing to a few principal type investigation. Alternatively, an ‘exposomic’ screen,
components that maximize orthogonality; weighted evaluating a large universe of environmental factors, can
quantile regression can estimate the effect of the whole be used in combination with a single or small set of
mixture on an outcome and apportion the relative con- genes. Both approaches require large samples—optimally,
tributions of the parts [Czarnota, Gennings, & Wheeler, thousands to tens of thousands of participants. The

20 Hertz-Picciotto et al./Environmental contributions to ASD INSAR


Licensed to Mariana Wollinger Maciel Fassheber - marianawmaciel@gmail.com - 046.925.309-60

primary limitation in either of these statistically equiva- epigenome to up- or down-regulate gene expression.
lent strategies is the sample size needs for reasonably Thus, as epidemiologists move into gene-environment
powered analysis of interactions involving combined interaction, epigenetics opens the door to understand-
effects, when either genes or environmental exposures ing the truly complex interplay between inherited and
are of high dimension (thousands to hundreds of thou- noninherited molecules within our bodies and our eco-
sands of variables), even when the other has only a few logic niche.
categories or even one binary or continuous factor.
Environmental Epigenetics
Examining two high-dimensional sets of data– both
genomic and exposomic—in combination is a feasibility DNA methylation, histone modifications, micro-RNA,
challenge. Practical limits for sample sizes of such omic- noncoding RNAs, and chromatin remodeling all consti-
by-omic analysis have been described [Patel, 2016]: tute changes in the architecture and expression of
under a wide range of assumptions, statistical power for genes, without affecting the DNA sequence itself. A bur-
detecting GxE interactions based on as few as 10 to geoning field of environmental epigenetics has identi-
over 10,000 environmental exposures, even with sam- fied epigenetic changes or profiles associated with a
ple sizes of 10,000 to 60,000, is low. Moreover, interac- wide range of exposures. Environmentally-induced epi-
tions with SNPs will magnify the nonindependence genetic changes, especially during periods of erasure
problem discussed above [Rao & Province, 2016]. and reestablishment of the methylome during preg-
Although greater power may be achieved, for example, nancy [Reik & Walter, 2001], have potential as both
for exposures more prevalent than the 20% maximum
markers and mediators of environmental influences on
assumed by Patel, or associated with greater interaction
later developmental outcomes [Baccarelli & Bollati,
relative risks [as was observed by Schmidt et al., 2011],
2009]. Human epidemiology and animal models indi-
alternative strategies for gene-by-environment analyses
cate that prenatal and postnatal environmental expo-
are needed.
sures influence developmental trajectories, leading to
Several alternatives to omic-omic analyses are feasible,
persistent changes in phenotypes and in chronic disease
and lie within reach even for current sample sizes,
susceptibility. Evidence is accumulating for epigenetic
including: biologically-based pathways relevant to spe-
effects, primarily DNA methylation in offspring, from
cific exposures [Schmidt et al., 2011, 2012; Volk et al.,
environmental exposures including stress, socioeco-
2014]; expanded gene pathways, for example, based on
nomic status, parental body mass index, gestational dia-
gene ontology and/or toxicogenomic databases, to
betes, maternal antibiotic use, lead, arsenic, mercury,
restrict the number of genes in combination with a
bisphenol A, cigarette smoke, pesticides, and nutritional
mechanistically defined limited set of candidate expo-
factors [Breton et al., 2017; Finer et al., 2015; Ladd-
sures; genome-wide searches in combination with candi-
Acosta et al., 2014; Schmidt et al., 2016]. Similarly,
date or known autism risk factors; and use of two-stage
methods for GWAS in combination with various new numerous studies have provided evidence for DNA
exposomic analyses. Variations on these approaches are methylation or other epigenetic alterations in ASD
discussed in [Patel et al., 2012; Patel, Chen, Kodama, [Feinberg et al., 2015; Schroeder, Lott, Korf, & LaSalle,
Ioannidis, & Butte, 2013; Patel, 2016]. The recently 2011; Schroeder et al., 2016] and other neurodevelop-
launched ECHO (Environmental Influences on Child mental disorders as reviewed previously [Flashner,
Health Outcomes) program encompassing 50,000 chil- Russo, Boileau, Leong, & Gallicano, 2013; Kubota &
dren can be an excellent platform for advancing Big Data Mochizuki, 2016; Loke, Hannan, & Craig, 2015; Vogel
analysis of gene-environment interactions. Ciernia & LaSalle, 2016; Zhubi, Cook, Guidotti, & Gray-
Finally, the type of gene-environment interaction dis- son, 2014]. These studies implicate large epigenetic con-
cussed above—investigating the combination of inher- tributions to ASD etiology, and suggest epigenetic
ited genes and environmental exposures for synergistic mechanisms as mediators of effects from environmental
effects that are greater than the sum of the individual factors. Nevertheless, to date, direct causal evidence
effects—is only one of several ways in which the two can linking environmental factors with ASD through DNA
interact. A second is the ability of environmental factors methylation or other epigenetic mechanisms remains
to alter DNA through either (a) de novo mutational limited.
events or (b) de novo replication defects resulting in Most environmental epigenetics currently focuses on
copy number variants, either deletions or duplications, DNA methylation, due to greater feasibility in large-scale
sometimes numbering in hundreds or more repeats of epidemiologic studies. Even so, numerous methodologi-
genetic code. A third type of gene-environment interac- cal issues complicate measurement and interpretation of
tion involves epigenetic alterations of the genetic struc- DNA methylation marks, especially in relation to a brain
ture (see below) as a result of environmental factors, disorder like autism, including: species specificity, tissue
both endogenous and exogenous, acting upon the specificity, and cell heterogeneity [Bakulski & Fallin,

INSAR Hertz-Picciotto et al./Environmental contributions to ASD 21


Licensed to Mariana Wollinger Maciel Fassheber - marianawmaciel@gmail.com - 046.925.309-60

2014]. For instance, different cell types have distinct epi- uncover additional preventive measures. Although the
genetic marks, and the specific composition of blood preconception and prenatal periods may be the most
cells shift in the presence of, for example, inflammation. influential, continued plasticity of the central nervous
Other issues concern laboratory and statistical methods system postnatally implies further opportunity during
[Breton et al., 2017]. Nevertheless, epigenetics has the first year or two, to mitigate the debilitating fea-
opened up the big black box of how genes and environ- tures and enhance the special giftedness of ASD. Our
ment play complementary roles in development and goal is simple: by studying environmental—hence mal-
health. The complex genetic architecture of ASD alters leable—factors that contribute to ASD, we amplify the
susceptibility to environmental agents, while noninher- potential for a productive and fulfilling life for all
ited factors influence gene expression in tissue specific children.
contexts, affecting not only the exposed organism, but
potentially also future generations.
References
Developing and applying new technologies for mea-
suring the exposome, generating mega-studies with Abrahams, B.S., & Geschwind, D.H. (2008). Advances in autism
broad and high-quality genome and exposome data to genetics: On the threshold of a new neurobiology. Nature
be analyzed using robust strategies, and finally, tying it Reviews Genetics, 9, 341–355.
together with deep characterization of the epigenome Allen, L.H. (2000). Anemia and iron deficiency: Effects on
will be heavy challenges – yet not as far-flung as they pregnancy outcome. American Journal of Clinical Nutri-
tion, 71, 1280S–1284S.
seem today. We are poised to dramatically advance our
American Psychiatric Association. (1994). Diagnostic and statis-
understanding of the etiology of ASD as never before.
tical manual of mental disorders. Fourth edition (DSM-IV).
Washington, DC: American Psychiatric Association.
Conclusion Anney, R.J., Kenny, E.M., O’Dushlaine, C., Yaspan, B.L.,
Parkhomenka, E., Buxbaum, J.D. . . . Gallaghar, L. (2011).
As noted, many factors reviewed here have associations Gene-ontology enrichment analysis in two independent
family-based samples highlights biologically plausible pro-
with other neurodevelopmental or psychiatric condi-
cesses for autism spectrum disorders. European Journal of
tions, and therefore may lack specificity for autism.
Human Genetics, 19, 1082–1089.
Genetic factors or critical time periods may influence Centers for Disease Control. (1991). Use of folic acid for pre-
how these xenobiotics or noninherited conditions alter vention of spina bifida and other neural tube defects–1983–
brain connectivity and whether they lead to autism ver- 1991. MMWR Morbidity & Mortality Weekly Report, 40,
sus different deficits. Thus, the causal concepts of tim- 513–516.
ing and susceptibility loom as keys to unlocking the Araujo, T.G., Oliveira, A.G., Carvalho, B.M., Guadagnini, D.,
mysteries of autism etiology. Realistic strategies for pur- Protzek, A.O., Carvalheira, J.B., . . . Saad, M.G. (2012). Hepa-
suit of gene-by-environment synergies have been out- tocyte growth factor plays a key role in insulin resistance-
lined and many of the tools are now available, but associated compensatory mechanisms. Endocrinology, 153,
require large enough studies. New developments in epi- 5760–5769.
Atladottir, H.O., Parner, E.T., Schendel, D., Dalsgaard, S.,
genetics offer yet further opportunities to capture the
Thomsen, P.H., & Thorsen, P. (2007). Variation in inci-
complexity of environmental influences on this disor-
dence of neurodevelopmental disorders with season of
der. Major research gaps include determination of criti-
birth. Epidemiology, 18, 240–245.
cal etiologic windows for environmental exposures, Atladottir, H.O., Thorsen, P., Ostergaard, L., Schendel, D.E.,
how these vary by type of exposure, and whether epige- Lemcke, S., Abdallah, M., . . . Parner, E.T. (2010). Maternal
netic vulnerabilities feature in defining those windows. infection requiring hospitalization during pregnancy and
Parent-of-origin effects and hormonal hypotheses may autism spectrum disorders. Journal of Autism and Develop-
enable disentangling the maternal and paternal contri- mental Disorder, 40, 1423–1430.
butions and the sex ratio. Additional work on how the Atladottir, H.O., Henriksen, T.B., Schendel, D.E., & Parner, E.T.
gut microbiotal populations influence behavior, (2012). Autism after infection, febrile episodes, and antibi-
whether probiotics can induce desired changes, and if otic use during pregnancy: An exploratory study. Pediatrics,
130, e1447–e1454.
so, in what time periods, is also needed. Continuing
Avella-Garcia, C.B., & Julvez, J. (2014). Seafood intake and
targeted investigations into environmental chemicals,
neurodevelopment: A systematic review. Current Environ-
particularly endocrine disrupting compounds, appears
mental Health Reports, 1, 46–77.
to be another promising avenue with realistic opportu- Baccarelli, A., & Bollati, V. (2009). Epigenetics and environ-
nities for interventions. Similarly, deeper dives into mental chemicals. Current Opinion in Pediatrics, 21, 243–
maternal nutritional, obstetric, metabolic and other fac- 251.
tors during the pre-conception, prenatal and perinatal Baduel, C., Mueller, J.F., Tsai, H., & Gomez Ramos, M.J.
periods are likely to enhance understanding and (2015). Development of sample extraction and clean-up

22 Hertz-Picciotto et al./Environmental contributions to ASD INSAR


Licensed to Mariana Wollinger Maciel Fassheber - marianawmaciel@gmail.com - 046.925.309-60

strategies for target and non-target analysis of environmen- Are autistic-behaviors in children related to prenatal vita-
tal contaminants in biological matrices. Journal of Chroma- min use and maternal whole blood folate concentrations?
tography A, 1426, 33–47. Journal of Autism and Developmental Disorders, 44, 2602–
Bailey, A., Le Couteur, A., Gottesman, I., Bolton, P., Simonoff, 2607.
E., Yuzda, E., . . . Rutter, M. (1995). Autism as a strongly Braun, J.M., Kalkbrenner, A.E., Just, A.C., Yolton, K., Calafat,
genetic disorder: Evidence from a British twin study. Psy- A.M., Sjodin, A., . . . Lanphear, B.P. (2014b). Gestational
chological Medicine, 25, 63–77. exposure to endocrine-disrupting chemicals and reciprocal
Bakulski, K.M., & Fallin, M.D. (2014). Epigenetic epidemiology: social, repetitive, and stereotypic behaviors in 4- and 5-
Promises for public health research. Environmental and year-old children: The home study. Environmental Health
Molecular Mutagenesis, 55, 171–183. Perspectives, 122, 513–520.
Bandodkar, A.J., & Wang, J. (2014). Non-invasive wearable Braun, J.M., Yolton, K., Stacy S.L., Erar, B., Papandonatos,
electrochemical sensors: A review. Trends in Biotechnology, G.D., Bellinger, D.C., . . . Chen, A. (2017). Prenatal environ-
32, 363–371. mental chemical exposures and longitudinal patterns of
Banks, C.N., & Lein, P.J. (2012). A review of experimental evi- child neurobehavior. Neurotoxicology 62, 192–199.
dence linking neurotoxic organophosphorus compounds Braunschweig, D., Duncanson, P., Boyce, R., Hansen, R.,
and inflammation. Neurotoxicology, 33, 575–584. Ashwood, P., Pessah, I.N., . . . Van de Water J. (2012).
Baron-Cohen, S., Auyeung, B., Norgaard-Pedersen, B., Behavioral correlates of maternal antibody status among
Hougaard, D.M., Abdallah, M.W., Melgaard, L., . . . children with autism. Journal of Autism and Developmen-
Lombardo, M.V. (2015). Elevated fetal steroidogenic activity tal Disorders, 42, 1435–1445.
in autism. Molecular Psychiatry, 20, 369–376. Braunschweig, D., Krakowiak, P., Duncanson, P., Boyce, R.,
Bartlett, C.W., Gharani, N., Millonig, J.H., & Brzustowicz, L.M. Hansen, R.L., Ashwood, P., . . . Van de Water J. (2013).
(2005). Three autism candidate genes: A synthesis of
Autism-specific maternal autoantibodies recognize critical
human genetic analysis with other disciplines. Interna-
proteins in developing brain. Translational Psychiatry, 3,
tional Journal of Developmental Neuroscience, 23, 221–
e277.
234.
Bray, G.A., Nielsen, S.J., & Popkin, B.M. (2004). Consumption
Beard, J.L. (2000). Effectiveness and strategies of iron supple-
of high-fructose corn syrup in beverages may play a role in
mentation during pregnancy. American Journal of Clinical
the epidemic of obesity. American Journal of Clinical Nutri-
Nutrition, 71, 1288S–1294S.
tion, 79, 537–543.
Becerra, T.A., Wilhelm, M., Olsen, J., Cockburn, M., & Ritz, B.
Breton, C.V., Marsit, C.J., Faustman, E., Nadeau, K., Goodrich,
(2013). Ambient air pollution and autism in Los Angeles
J.M., Dolinoy, D.C., . . . Murphy S.K. (2017). Small-magni-
county, California. Environmental Health Perspectives, 121,
tude effect sizes in epigenetic end points are important in
380–386.
children’s environmental health studies: The children’s
Bespalova, I.N., & Buxbaum, J.D. (2003). Disease susceptibility
environmental health and disease prevention research cen-
genes for autism. Annals of Medicine, 35, 274–281.
ter’s epigenetics working group. Environmental Health Per-
Bettelheim, B. (1967). The empty fortress: Infantile autism and
spectives, 125, 511–526.
the birth of the self. New York, NY: Free Press.
Brigandi, S.A., Shao, H., Qian, S.Y., Shen, Y., Wu, B.L., & Kang,
Beversdorf, D.Q., Manning, S.E., Hillier, A., Anderson, S.L.,
J.X. (2015). Autistic children exhibit decreased levels of
Nordgren, R.E., Walters, S.E., . . . Bauman, M.L. (2005). Tim-
ing of prenatal stressors and autism. Journal of Autism and essential fatty acids in red blood cells. International Journal
Developmental Disorders, 35, 471–478. of Molecular Sciences, 16, 10061–10076.
Bleich, S.N., Wang, Y.C., Wang, Y., & Gortmaker, S.L. (2009). Buchmayer, S., Johansson, S., Johansson, A., Hultman, C.M.,
Increasing consumption of sugar-sweetened beverages Sparen, P., & Cnattingius, S. (2009). Can association
among us adults: 1988–1994 to 1999–2004. American Jour- between preterm birth and autism be explained by mater-
nal of Clinical Nutrition, 89, 372–381. nal or neonatal morbidity? Pediatrics, 124, e817–e825.
Blom, H.J., Shaw, G.M., den Heijer, M., & Finnell, R.H. (2006). Burns, C.J., Cohen, S.Z., & Lunchick, C. (2015). Neurodevelop-
Neural tube defects and folate: Case far from closed. Nature mental disorders and agricultural pesticide exposures. Envi-
Reviews Neuroscience, 7, 724–731. ronmental Health Perspectives, 123, A79.
Blom, H.J. (2009). Folic acid, methylation and neural tube clo- Burstyn, I., Sithole, F., & Zwaigenbaum, L. (2010). Autism spec-
sure in humans. Birth Defects Research Part A Clinical and trum disorders, maternal characteristics and obstetric com-
Molecular Teratology, 85, 295–302. plications among singletons born in Alberta, Canada.
Boitard, M., Bocchi, R., Egervari, K., Petrenko, V., Viale, B., Chronic Diseases in Canada, 30, 125–134.
Gremaud, S., . . . Kiss, J.Z. (2015). Wnt signaling regulates Burstyn, I., Wang, X., Yasui, Y., Sithole, F., & Zwaigenbaum, L.
multipolar-to-bipolar transition of migrating neurons in the (2011). Autism spectrum disorders and fetal hypoxia in a
cerebral cortex. Cell Reports, 10, 1349–1361. population-based cohort: Accounting for missing exposures
Bouchard, M.F., Chevrier, J., Harley, K.G., Kogut, K., Vedar, via estimation-maximization algorithm. BMC Medical
M., Calderon, N., . . . Eskenazi, B. (2011). Prenatal exposure Research Methodology, 11, 2.
to organophosphate pesticides and IQ in 7-year-old chil- Campbell, D.B., Li, C., Sutcliffe, J.S., Persico, A.M., & Levitt, P.
dren. Environmental Health Perspectives, 119, 1189–1195. (2008). Genetic evidence implicating multiple genes in the
Braun, J.M., Froehlich, T., Kalkbrenner, A., Pfeiffer, C.M., met receptor tyrosine kinase pathway in autism spectrum
Fazili, Z., Yolton, K., . . . Lanphear, B.P. (2014a). Brief report: disorder. Autism Research, 1, 159–168.

INSAR Hertz-Picciotto et al./Environmental contributions to ASD 23


Licensed to Mariana Wollinger Maciel Fassheber - marianawmaciel@gmail.com - 046.925.309-60

Centers for Disease Control and Prevention. (2016). National Coo, H., Ouellette-Kuntz, H., Lam, Y.M., Brownell, M., Flavin,
biomonitoring program fact sheet: Phthalates. https://www. M.P., & Roos, L.L. (2015). The association between the
cdc.gov/biomonitoring/phthalates_factsheet.html. interpregnancy interval and autism spectrum disorder in a
Chaidez, V., Hansen, R.L., & Hertz-Picciotto, I. (2014). Gastro- Canadian cohort. Canadian Journal of Public Health, 106,
intestinal problems in children with autism, developmental e36–e42.
delays or typical development. Journal of Autism Develop- Cooney, C.A., Dave, A.A., & Wolff, G.L. (2002). Maternal
mental Disorder, 44, 1117–1127. methyl supplements in mice affect epigenetic variation and
Chauhan, A., Chauhan, V., Brown, W.T., & Cohen, I. (2004). DNA methylation of offspring. Journal of Nutrition, 132,
Oxidative stress in autism: Increased lipid peroxidation and 2393S–2400S.
reduced serum levels of ceruloplasmin and transferrin–the Courchesne, E., Yeung-Courchesne, R., Press, G.A., Hesselink,
antioxidant proteins. Life Sciences, 75, 2539–2549. J.R., & Jernigan, T.L. (1988). Hypoplasia of cerebellar vermal
Chen, H., Xue, M., Mei, Z., Bambang Oetomo, S., & Chen, W. lobules VI and VII in autism. New England Journal of Medi-
(2016). A review of wearable sensor systems for monitoring cine, 318, 1349–1354.
body movements of neonates. Sensors, 16, E2134 Croen, L.A., Grether, J.K., Yoshida, C.K., Odouli, R., &
Chen, Y.J., & Hsu, C.C. (1994). Effects of prenatal exposure to Hendrick, V. (2011). Antidepressant use during pregnancy
PCBs on the neurological function of children: A neuropsy- and childhood autism spectrum disorders. Archives of Gen-
chological and neurophysiological study. Developmental eral Psychiatry, 68, 1104–1112.
Medicine and Child Neurology, 36, 312–320. Czarnota, J., Gennings, C., & Wheeler, D.C. (2015). Assessment
Cherta, L., Portoles, T., Pitarch, E., Beltran, J., Lopez, F.J., of weighted quantile sum regression for modeling chemical
ndez, F. (2015). Analytical strategy
Calatayud, C., . . . Herna mixtures and cancer risk. Cancer Informatics, 14, 159–171.
based on the combination of gas chromatography coupled Davis, E.P., & Pfaff, D. (2014). Sexually dimorphic responses to
to time-of-flight and hybrid quadrupole time-of-flight mass early adversity: Implications for affective problems and
analyzers for non-target analysis in food packaging. Food autism spectrum disorder. Psychoneuroendocrinology, 49,
Chemistry, 188, 301–308. 11–25.
Cheslack-Postava, K., Liu, K., & Bearman, P.S. (2011). Closely De Angelis, M., Francavilla, R., Piccolo, M., De Giacomo, A., &
spaced pregnancies are associated with increased odds of Gobbetti, M. (2015). Autism spectrum disorders and intesti-
autism in California sibling births. Pediatrics, 127, 246–253. nal microbiota. Gut Microbes, 6, 207–213.
Cheslack-Postava, K., Rantakokko, P.V., Hinkka-Yli-Salomaki, De Rubeis, S., He, X., Goldberg, A.P., Poultney, C.S., Samocha,
S., Surcel, H.M., McKeague, I.W., Kiviranta, H.A., . . . Brown K., Cicek, A.E., . . . Buxbaum, J.D. (2014). Synaptic, tran-
A.S. (2013). Maternal serum persistent organic pollutants in scriptional and chromatin genes disrupted in autism.
the finnish prenatal study of autism: A pilot study. Neuro- Nature, 515, 209–215.
toxicology and Teratology, 38, 1–5. Demirci, C., Ernst, S., Alvarez-Perez, J.C., Rosa, T., Valle, S.,
Cheslack-Postava, K., Suominen, A., Jokiranta, E., Lehti, V., Shridhar, V., . . . Garcıa-Ocana, A. (2012). Loss of HGF/c-met
McKeague, I.W., Sourander, A., . . . Brown A.S. (2014). signaling in pancreatic beta-cells leads to incomplete mater-
Increased risk of autism spectrum disorders at short and nal beta-cell adaptation and gestational diabetes mellitus.
long interpregnancy intervals in Finland. Journal of Ameri- Diabetes, 61, 1143–1152.
can Academy of Child and Adolescent Psychiatry, 53, Desmond, M.M., Wilson, G.S., Melnick, J.L., Singer, D.B., Zion,
1074–1081.e4. T.E., Rudolph, A.J., . . . Blattner, R.J. (1967). Congenital
Chess, S. (1977). Follow-up report on autism in congenital rubella encephalitis. Course and early sequelae. Journal of
rubella. Journal of Autism and Childhood Schizophrenia, 7, Pediatrics, 71, 311–331.
69–81. Deykin, E.Y., & MacMahon, B. (1979). Viral exposure and
Cohen, I.L., Sudhalter, V., Pfadt, A., Jenkins, E.C., Brown, autism. American Journal of Epidemiology, 109, 628–638.
W.T., & Vietze, P.M. (1991). Why are autism and the Dietz, P.M., Homa, D., England, L.J., Burley, K., Tong, V.T.,
fragile-x syndrome associated? Conceptual and methodo- Dube, S.R., . . . Bernert, J.T. (2011). Estimates of nondisclo-
logical issues. American Journal of Human Genetics, 48, sure of cigarette smoking among pregnant and nonpreg-
195–202. nant women of reproductive age in the United States.
Colomer, J., Colomer, C., Gutierrez, D., Jubert, A., Nolasco, A., American Journal of Epidemiology, 173, 355–359.
Donat, J., . . . Alvarez-Dardet, C. (1990). Anaemia during Dodds, L., Fell, D.B., Shea, S., Armson, B.A., Allen, A.C., &
pregnancy as a risk factor for infant iron deficiency: Report Bryson, S. (2011). The role of prenatal, obstetric and neona-
from the valencia infant anaemia cohort (VIAC) study. Pae- tal factors in the development of autism. Journal of Autism
diatric and Perinatal Epidemiology, 4, 196–204. and Developmental Disorders, 41, 891–902.
Conde-Agudelo, A., Rosas-Bermudez, A., & Norton, M.H. Dolinoy, D.C., Huang, D., & Jirtle, R.L. (2007). Maternal nutri-
(2016). Birth spacing and risk of autism and other neurode- ent supplementation counteracts bisphenol a-induced DNA
velopmental disabilities: A systematic review. Pediatrics, hypomethylation in early development. Proceedings of the
137, National Academy of Sciences of the United States of Amer-
Connolly, N., Anixt, J., Manning, P., Ping, I.L.D., Marsolo, ica, 104, 13056–13061.
K.A., & Bowers, K. (2016). Maternal metabolic risk factors Durkin, M.S., DuBois, L.A., & Maenner, M.J. (2015). Inter-preg-
for autism spectrum disorder-an analysis of electronic medi- nancy intervals and the risk of autism spectrum disorder:
cal records and linked birth data. Autism Research, 9, 829– Results of a population-based study. Journal of Autism and
837. Developmental Disorders, 45, 2056–2066.

24 Hertz-Picciotto et al./Environmental contributions to ASD INSAR


Licensed to Mariana Wollinger Maciel Fassheber - marianawmaciel@gmail.com - 046.925.309-60

Duthie, S.J. (1999). Folic acid deficiency and cancer: Mecha- Folstein, S.E., & Rutter, M.L. (1988). Autism: Familial aggrega-
nisms of DNA instability. British Medical Bulletin, 55, 578– tion and genetic implications. Journal of Autism and Devel-
592. opmental Disorders, 18, 3–30.
Eidelman, A.I., & Samueloff, A. (2002). The pathophysiology Folstein, S.E., & Piven, J. (1991). Etiology of autism: Genetic
of the fetus of the diabetic mother. Seminars in Perinatol- influences. Pediatrics, 87, 767–773.
ogy, 26, 232–236. Freeman, M.P., Hibbeln, J.R., Wisner, K.L., Davis, J.M.,
El Marroun, H., White, T.J., van der Knaap, N.J., Homberg, Mischoulon, D., Peet, M., . . . Stoll, A.L. (2006). Omega-3
J.R., Fernandez, G., Schoemaker, N.K., . . . Teimeire, H. fatty acids: Evidence basis for treatment and future research
(2014). Prenatal exposure to selective serotonin reuptake in psychiatry. The Journal of Clinical Psychiatry, 67, 1954–
inhibitors and social responsiveness symptoms of autism: 1967.
Population-based study of young children. British Journal Gadow, K.D., Perlman, G., Ramdhany, L., & de Ruiter, J.
of Psychiatry, 205, 95–102. (2016). Clinical correlates of co-occurring psychiatric and
Engel, S.M., Miodovnik, A., Canfield, R.L., Zhu, C., Silva, M.J., autism spectrum disorder (ASD) symptom-induced impair-
Calafat, A.M., . . . Wolff, M.S. (2010). Prenatal phthalate ment in children with ASD. Journal of Abnormal Child Psy-
exposure is associated with childhood behavior and execu- chology, 44, 129–139.
tive functioning. Environmental Health Perspectives, 118, Gardener, H., Spiegelman, D., & Buka, S.L. (2009). Prenatal risk
565–571. factors for autism: Comprehensive meta-analysis. British
Engel, S.M., Wetmur, J., Chen, J., Zhu, C., Barr, D.B., Canfield, Journal of Psychiatry, 195, 7–14.
R.L., . . . Wolff, M.S. (2011). Prenatal exposure to organo- Gaugler, T., Klei, L., Sanders, S.J., Bodea, C.A., Goldberg, A.P.,
phosphates, paraoxonase 1, and cognitive development in Lee, A.B., . . . Buxbaum, J.D. (2014). Most genetic risk for
childhood. Environmental Health Perspectives, 119, 1182– autism resides with common variation. Nature Genetics,
1188.
46, 881–885.
Ervin, R.B. (2009). Prevalence of metabolic syndrome among
Geiss, L.S., Wang, J., Cheng, Y.J., Thompson, T.J., Barker, L.,
adults 20 years of age and over, by sex, age, race and eth-
Li, Y., . . . Gregg, E.W. (2014). Prevalence and incidence
nicity, and body mass index: United states, 2003–2006.
trends for diagnosed diabetes among adults aged 20 to 79
National Health Statistics Reports, 1–7.
years, united states, 1980–2012. JAMA, 312, 1218–1226.
Eskenazi, B., Marks, A.R., Bradman, A., Harley, K., Barr, D.B.,
Gesundheit, B., Rosenzweig, J.P., Naor, D., Lerer, B., Zachor,
Johnson, C., . . . Jewell, N.P. (2007). Organophosphate pesti-
D.A., Prochazka, V., . . . Ashwood, P. (2013). Immunological
cide exposure and neurodevelopment in young Mexican-
and autoimmune considerations of autism spectrum disor-
American children. Environmental Health Perspectives,
ders. Journal of Autoimmunity, 44, 1–7.
115, 792–798.
Gibson, J.L. (1904). A plea for painted railings on painted walls
Feinberg, J.I., Bakulski, K.M., Jaffe, A.E., Tryggvadottir, R.,
of rooms as the source of lead poisoning amongst Queens-
Brown, S.C., Goldman, L.R., . . . Feinburg, A.P. (2015). Pater-
land children. Australian Medical Gazette, 23, 149–153.
nal sperm DNA methylation associated with early signs of
Reprinted in Public Health Reports 2005 (May-June);120:
autism risk in an autism-enriched cohort. International
301–304.
Journal of Epidemiology, 44, 1199–1210.
Gong, T., Dalman, C., Wicks, S., Dal, H., Magnusson, C.,
Finegold, S.M., Dowd, S.E., Gontcharova, V., Liu, C., Henley,
Lundholm, C., . . . Pershagen, G. (2017). Perinatal exposure
K.E., Wolcott, R.D., . . . Green, J.A. (2010). Pyrosequencing
study of fecal microflora of autistic and control children. to traffic-related air pollution and autism spectrum disor-
Anaerobe, 16, 444–453. ders. Environmental Health Perspectives, 125, 119–126.
Finer, S., Mathews, C., Lowe, R., Smart, M., Hillman, S., Foo, Gouma, P.I., Wang, L., Simon, S.R., & Stanacevic, M. (2017).
L., . . . Hitman, G.A. (2015). Maternal gestational diabetes is Novel isoprene sensor for a flu virus breath monitor. Sen-
associated with genome-wide DNA methylation variation in sors, 17,
placenta and cord blood of exposed offspring. Human Grether, J.K., Li, S.X., Yoshida, C.K., & Croen, L.A. (2010).
Molecular Genetics, 24, 3021–3029. Antenatal ultrasound and risk of autism spectrum disorders.
Flashner, B.M., Russo, M.E., Boileau, J.E., Leong, D.W., & Journal of Autism and Developmental Disorders, 40, 238–
Gallicano, G.I. (2013). Epigenetic factors and autism spec- 245.
trum disorders. Neuromolecular Medicine, 15, 339–350. Gunnes, N., Suren, P., Bresnahan, M., Hornig, M., Lie, K.K.,
Flegal, K.M., Carroll, M.D., Ogden, C.L., & Curtin, L.R. (2010). Lipkin, W.I., . . . Stoltenberg, C. (2013). Interpregnancy
Prevalence and trends in obesity among us adults, 1999– interval and risk of autistic disorder. Epidemiology, 24,
2008. JAMA, 303, 235–241. 906–912.
Flegal, K.M., Carroll, M.D., Kit, B.K., & Ogden, C.L. (2012). Guxens, M., Ghassabian, A., Gong, T., Garcia-Esteban, R.,
Prevalence of obesity and trends in the distribution of body Porta, D., Giorgis-Allemand, L., . . . Jordi, S. (2016). Air pol-
mass index among us adults, 1999–2010. JAMA, 307, 491– lution exposure during pregnancy and childhood autistic
497. traits in four European population-based cohort studies:
Folstein, S., & Rutter, M. (1977a). Genetic influences and The escape project. Environmental Health Perspectives,
infantile autism. Nature, 265, 726–728. 124, 133–140.
Folstein, S., & Rutter, M. (1977b). Infantile autism: A genetic Hammel, S.C., Hoffman, K., Webster, T.F., Anderson, K.A., &
study of 21 twin pairs. Journal of Child Psychology and Stapleton, H.M. (2016). Correction to measuring personal
Psychiatry, 18, 297–321. exposure to organophosphate flame retardants using

INSAR Hertz-Picciotto et al./Environmental contributions to ASD 25


Licensed to Mariana Wollinger Maciel Fassheber - marianawmaciel@gmail.com - 046.925.309-60

silicone wristbands and hand wipes. Environmental Science (pfoa) in communities surrounding a fluoropolymer pro-
and Technology, 50, 10291. duction facility. Environmental Health Perspectives, 119,
Harada, M. (1978). Congenital minamata disease: Intrauterine 92–97.
methylmercury poisoning. Teratology, 18, 285–288. Hornig, M., Bresnahan, M.A., Che, X., Schultz, A.F., Ukaigwe,
Harada, M. (1995). Minamata disease: Methylmercury poison- J.E., Eddy, M.L., . . . Lipkin, W.I. (2017). Prenatal fever and
ing in japan caused by environmental pollution. Critical autism risk. Molecular Psychiatry. doi: 10.1038/
Reviews in Toxicology, 25, 1–24. mp.2017.119.
Harrington, R.A., Lee, L.C., Crum, R.M., Zimmerman, A.W., & Howe, C.G., Niedzwiecki, M.M., Hall, M.N., Liu, X., Ilievski,
Hertz-Picciotto, I. (2014). Prenatal SSRI use and offspring V., Slavkovich, V., . . . Gamble, M.V. (2014). Folate
with autism spectrum disorder or developmental delay. and cobalamin modify associations between s-
Pediatrics, 133, e1241–e1248. adenosylmethionine and methylated arsenic metabolites in
Hartz, S.C., Heinonen, O.P., Shapiro, S., Siskind, V., & Slone, arsenic-exposed Bangladeshi adults. Journal of Nutrition,
D. (1975). Antenatal exposure to meprobamate and chlordi- 144, 690–697.
azepoxide in relation to malformations, mental develop- Hsiao, E.Y. (2013). Immune dysregulation in autism spectrum
ment, and childhood mortality. New England Journal of disorder. International Review of Neurobiology, 113, 269–
Medicine, 292, 726–728. 302.
Hebert, K.J., Miller, L.L., & Joinson, C.J. (2010). Association of Hsu, S.T., Ma, C.I., Hsu, S.K., Wu, S.S., Hsu, N.H., Yeh, C.C.,
autistic spectrum disorder with season of birth and concep- . . . Wu, S.B. (1985). Discovery and epidemiology of PCB
tion in a UK cohort. Autism Research, 3, 185–190. poisoning in Taiwan: A four-year follow-up. Environmental
Heijmans, B.T., Tobi, E.W., Stein, A.D., Putter, H., Blauw, G.J., Health Perspectives, 59, 5–10.
Susser, E.S., . . . Lumey, L.H. (2008). Persistent epigenetic dif- Hultman, C.M., Sparen, P., & Cnattingius, S. (2002). Perinatal
ferences associated with prenatal exposure to famine in
risk factors for infantile autism. Epidemiology, 13, 417–423.
humans. Proceedings of the National Academy of Sciences
Hviid, A., Melbye, M., & Pasternak B. (2013). Use of selective
of the United States of America, 105, 17046–17049.
serotonin reuptake inhibitors during pregnancy and risk of
Herbert, M., & Weintraub, K. 2012. The autism revolution.
autism. N Engl J Med 369(25), 2406–2415
New York, NY: Ballantine Books.
Jackson, P.B., Boccuto, L., Skinner, C., Collins, J.S., Neri, G.,
Herbstman, J.B., Sjodin, A., Kurzon, M., Lederman, S.A., Jones,
Gurrieri, F., . . . Schwartz, C.E. (2009). Further evidence that
R.S., Rauh, V., . . . Perera, F.. (2010). Prenatal exposure to
the rs1858830 c variant in the promoter region of the met
PBDEs and neurodevelopment. Environmental Health Per-
gene is associated with autistic disorder. Autism Research,
spectives, 118, 712–719.
2, 232–236.
Hertz-Picciotto, I., Pastore, L.M., & Beaumont, J.J. (1996). Tim-
Jacobson, J.L., & Jacobson, S.W. (1996). Dose-response in peri-
ing and patterns of exposures during pregnancy and their
natal exposure to polychlorinated biphenyls (PCBs): The
implications for study methods. American Journal of Epide-
Michigan and North Carolina cohort studies. Toxicology
miology, 143, 597–607.
and Industrial Health, 12, 435–445.
Hertz-Picciotto, I., Croen, L.A., Hansen, R., Jones, C.R., van de
James, S.J., Cutler, P., Melnyk, S., Jernigan, S., Janak, L.,
Water, J., & Pessah, I.N. (2006). The charge study: An epide-
Gaylor, D.W., . . . Neubrander, J.A. (2004). Metabolic bio-
miologic investigation of genetic and environmental factors
markers of increased oxidative stress and impaired methyla-
contributing to autism. Environmental Health Perspectives,
114, 1119–1125. tion capacity in children with autism. American Journal of
Hertz-Picciotto, I., & Delwiche, L. (2009a). The rise in autism Clinical Nutrition, 80, 1611–1617.
and the role of age at diagnosis. Epidemiology, 20, 84–90. James, S.J., Melnyk, S., Fuchs, G., Reid, T., Jernigan, S., Pavliv,
Hertz-Picciotto, I., & Delwiche, L. (2009b). Estimating the inci- O., . . . Gaylor, D.W. (2009). Efficacy of methylcobalamin
dence of autism. Epidemiology, 20, 623–624. 610.1097/ and folinic acid treatment on glutathione redox status in
EDE.1090b1013e3181a1081ef1099. children with autism. American Journal of Clinical Nutri-
Heuer, L., Braunschweig, D., Ashwood, P., Van de Water, J., & tion, 89, 425–430.
Campbell, D.B. (2011). Association of a met genetic variant Jolous-Jamshidi, B., Cromwell, H.C., McFarland A.M., &
with autism-associated maternal autoantibodies to fetal Meserve L.A. (2010). Perinatal exposure to polychlorinated
brain proteins and cytokine expression. Translational Psy- biphenyls alters social behaviors in rats. Toxicol Lett
chiatry, 1, e48. 199(2), 136–143.
Hill, A.B. (1965). The environment and disease: Association or Jory, J. (2016). Abnormal fatty acids in Canadian children with
causation? Proceedings of the Royal Society of Medicine, autism. Nutrition, 32, 474–477.
58, 295–300. Joshi, R., Adhikari, S., Patro, B.S., Chattopadhyay, S., &
Hoffman, K., Webster, T.F., Weisskopf, M.G., Weinberg, J., & Mukherjee, T. (2001). Free radical scavenging behavior of
Vieira, V.M. (2010). Exposure to polyfluoroalkyl chemicals folic acid: Evidence for possible antioxidant activity. Free
and attention deficit/hyperactivity disorder in U.S. Children Radical Biology and Medicine, 30, 1390–1399.
12–15 years of age. Environmental Health Perspectives, 118, Julvez, J., Fortuny, J., Mendez, M., Torrent, M., Ribas-Fito, N.,
1762–1767. & Sunyer, J. (2009). Maternal use of folic acid supplements
Hoffman, K., Webster, T.F., Bartell, S.M., Weisskopf, M.G., during pregnancy and four-year-old neurodevelopment in a
Fletcher, T., & Vieira, V.M. (2011). Private drinking water population-based birth cohort. Paediatric and Perinatal Epi-
wells as a source of exposure to perfluorooctanoic acid demiology, 23, 199–206.

26 Hertz-Picciotto et al./Environmental contributions to ASD INSAR


Licensed to Mariana Wollinger Maciel Fassheber - marianawmaciel@gmail.com - 046.925.309-60

Julvez, J., Mendez, M., Fernandez-Barres, S., Romaguera, D., brand of polychlorinated biphenyls. Environmental Health
Vioque, J., Llop, S., . . . Sunyer, J. (2016). Maternal consump- Perspectives, 1, 119–128.
tion of seafood in pregnancy and child neuropsychological Ladd-Acosta, C., Lee, B.K., Andrews, S.V., Newschaffer, C.J.,
development: A longitudinal study based on a population Schieve, L.A., Windham, G.C., . . . Fallin, M.D. (2014). DNA
with high consumption levels. American Journal of Epide- methylation as a biomarker for prenatal exposures impli-
miology, 183, 169–182. cated in autism spectrum disorders. Atlanta, GA: Interna-
Kalkbrenner, A.E., Daniels, J.L., Chen, J.C., Poole, C., Emch, tional Meeting for Autism Research.
M., & Morrissey, J. (2010). Perinatal exposure to hazardous Lambrou, A., Baccarelli, A., Wright, R.O., Weisskopf, M.,
air pollutants and autism spectrum disorders at age 8. Epi- Bollati, V., Amarasiriwardena, C., . . . Schwartz, J. (2012).
demiology, 21, 631–641. Arsenic exposure and DNA methylation among elderly
Kalkbrenner, A.E., Schmidt, R.J., & Penlesky, A.C. (2014). Envi- men. Epidemiology, 23, 668–676.
ronmental chemical exposures and autism spectrum disor- Larsson, M., Weiss, B., Janson, S., Sundell, J., & Bornehag, C.G.
ders: A review of the epidemiological evidence. Current (2009). Associations between indoor environmental factors
Problems in Pediatric and Adolescent Health Care, 44, 277– and parental-reported autistic spectrum disorders in chil-
318. dren 6–8 years of age. Neurotoxicology, 30, 822–831.
Kang, H.G., Kwon, K.H., Lee, I.W., Jung, B., Park, E.C., & Jang, Lawrence, J.M., Contreras, R., Chen, W., & Sacks, D.A. (2008).
S.I. (2013). Biochemically-verified smoking rate trends and Trends in the prevalence of preexisting diabetes and gesta-
factors associated with inaccurate self-reporting of smoking tional diabetes mellitus among a racially/ethnically diverse
habits in Korean women. Asian Pacific Journal of Cancer population of pregnant women, 1999–2005. Diabetes Care,
Prevention, 14, 6807–6812. 31, 899–904.
Karagas, M.R., Choi, A.L., Oken, E., Horvat, M., Schoeny, R., Lee, L.C., Newschaffer, C.J., Lessler, J.T., Lee, B.K., Shah, R., &
Kamai, E., . . . Korrick, S. (2012). Evidence on the human Zimmerman, A.W. (2008). Variation in season of birth in
health effects of low-level methylmercury exposure. Envi- singleton and multiple births concordant for autism spec-
ronmental Health Perspectives, 120, 799–806. trum disorders. Paediatric and Perinatal Epidemiology, 22,
Kim, S., Hong, S.H., Bong, C.K., & Cho, M.H. (2015). Charac- 172–179.
terization of air freshener emission: The potential health Li, M., Fallin, M.D., Riley, A., Landa, R., Walker, S.O.,
effects. Journal of Toxicological Sciences, 40, 535–550. Silverstein, M., . . . Wang, X. (2016). The association of
King, M., & Bearman, P. (2009). Diagnostic change and the maternal obesity and diabetes with autism and other devel-
increased prevalence of autism. Int J Epidemiol, 38, 1224– opmental disabilities. Pediatrics, 137, 1–10.
1234. Li, M.X., Yeung, J.M., Cherny, S.S., & Sham, P.C. (2012). Evalu-
Krakowiak, P., Walker, C.K., Bremer, A.A., Baker, A.S., Ozonoff, ating the effective numbers of independent tests and signif-
S., Hansen, R.L., . . . Hertz-Picciotto I. (2012). Maternal met- icant p-value thresholds in commercial genotyping arrays
abolic conditions and risk for autism and other neurodeve- and public imputation reference datasets. Human Genetics,
lopmental disorders. Pediatrics, 129, e1121–e1128. 131, 747–756.
Krakowiak, P., Walker, C.K., Tancredi, D.J., & Hertz-Picciotto, Loke, Y.J., Hannan, A.J., & Craig, J.M. (2015). The role of epi-
I. (2015). Maternal recall versus medical records of meta- genetic change in autism spectrum disorders. Frontiers in
bolic conditions from the prenatal period: A validation Neurology, 6, 107.
study. Maternal and Child Health Journal, 19, 1925–1935. Lopez-Espinosa, M.J., Mondal, D., Armstrong, B.G., Eskenazi,
Krakowiak, P., Goines, P.E., Tancredi, D.J., Ashwood, P., B., & Fletcher, T. (2016). Perfluoroalkyl substances, sex hor-
Hansen, R.L., Hertz-Picciotto, I., . . . Van de Water J. mones, and insulin-like growth factor-1 at 6–9 years of age:
(2017a). Neonatal cytokine profiles associated with autism A cross-sectional analysis within the c8 health project.
spectrum disorder. Biological Psychiatry, 81, 442–451. Environmental Health Perspectives, 124, 1269–1275.
Krakowiak, P., Walker, C.K., Tancredi, D., Hertz-Picciotto, I., & Lorber, M. (2008). Exposure of Americans to polybrominated
Van de Water, J. (2017b). Autism-specific maternal anti- diphenyl ethers. Journal of Exposure Science & Environ-
fetal brain autoantibodies are associated with metabolic mental Epidemiology, 18, 2–19.
conditions. Autism Research, 10, 89–98. Lyall, K., Pauls, D.L., Spiegelman, D., Ascherio, A., &
Kubota, T., & Mochizuki, K. (2016). Epigenetic effect of envi- Santangelo S.L. (2012). Pregnancy complications and
ronmental factors on autism spectrum disorders. Interna- obstetric suboptimality in association with autism spectrum
tional Journal of Environmental Research and Public disorders in children of the Nurses’ Health Study II. Autism
Health, 13, Res 5, 21–30.
Kung, K.T., Spencer, D., Pasterski, V., Neufeld, S., Glover, V., Lyall, K., Munger, K.L., O’Reilly, E.J., Santangelo, S.L., &
O’Connor, T.G., . . . Hines M. (2016). No relationship Ascherio, A. (2013). Maternal dietary fat intake in associa-
between prenatal androgen exposure and autistic traits: tion with autism spectrum disorders. American Journal of
Convergent evidence from studies of children with congen- Epidemiology, 178, 209–220.
ital adrenal hyperplasia and of amniotic testosterone con- Lyall, K., Schmidt, R.J., & Hertz-Picciotto, I. (2014). Maternal
centrations in typically developing children. Journal of lifestyle and environmental risk factors for autism spectrum
Child Psychology and Psychiatry, 57, 1455–1462. disorders. International Journal of Epidemiology, 43, 443–
Kuratsune, M., Yoshimura, T., Matsuzaka, J., & Yamaguchi, A. 464.
(1972). Epidemiologic study on yusho, a poisoning caused Lyall, K., Croen, L., Daniels, J., Fallin, M.D., Ladd-Acosta, C.,
by ingestion of rice oil contaminated with a commercial Lee, B.K., . . . Newschaffer, C. (2017a). The changing

INSAR Hertz-Picciotto et al./Environmental contributions to ASD 27


Licensed to Mariana Wollinger Maciel Fassheber - marianawmaciel@gmail.com - 046.925.309-60

epidemiology of autism spectrum disorders. Annual Review Martinez Bueno, M.J., Ulaszewska, M.M., Gomez, M.J.,
of Statistics and its Application, 38, 81–102. Hernando, M.D., & Fernandez-Alba, A.R. (2012). Simulta-
Lyall, K., Croen, L.A., Sjodin, A., Yoshida, C.K., Zerbo, O., neous measurement in mass and mass/mass mode for accu-
Kharrazi, M., . . . Windham, G.C. (2017b). Polychlorinated rate qualitative and quantitative screening analysis of
biphenyl and organochlorine pesticide concentrations in pharmaceuticals in river water. Journal of Chromatography
maternal mid-pregnancy serum samples: Association with A, 1256, 80–88.
autism spectrum disorder and intellectual disability. Envi- Mazahery, H., Stonehouse, W., Delshad, M., Kruger, M.C.,
ronmental Health Perspectives, 125, 474–480. Conlon, C.A., Beck, K.L., . . . von hurst, P.R. (2017). Rela-
Mackay, D.F., Smith, G.C., Cooper, S.A., Wood, R., King, A., tionship between long chain n-3 polyunsaturated fatty
Clark, D.N., . . . Pell, J.P. (2016). Month of conception and acids and autism spectrum disorder: Systematic review and
learning disabilities: A record-linkage study of 801,592 chil- meta-analysis of case-control and randomized controlled
dren. American Journal of Epidemiology, 184, 485–493. trials. Nutrients, 9, E155
Mahfouz, A., Ziats, M.N., Rennert, O.M., Lelieveldt, B.P., & Mazina, V., Gerdts, J., Trinh, S., Ankenman, K., Ward, T.,
Reinders, M.J. (2015). Shared pathways among autism can- Dennis, M.Y., . . . Bernier, R. (2015). Epigenetics of autism-
didate genes determined by co-expression network analysis related impairment: Copy number variation and maternal
of the developing human brain transcriptome. Journal of infection. Journal of Developmental and Behavioral Pediat-
Molecular Neuroscience, 57, 580–594. rics, 36, 61–67.
Mahic, M., Mjaaland, S., Bovelstad, H.M., Gunnes, N., Susser, McCarthy, M.M., & Wright, C.L. (2017). Convergence of sex
E., Bresnahan, M., . . . Lipkin, W.I. (2017). Maternal immu- differences and the neuroimmune system in autism spec-
noreactivity to herpes simplex virus 2 and risk of autism trum disorder. Biological Psychiatry, 81, 402–410.
spectrum disorder in male offspring. mSphere, 2, McNamara, R.K., & Carlson, S.E. (2006). Role of omega-3 fatty
Main, K.M., Mortensen, G.K., Kaleva, M.M., Boisen, K.A., acids in brain development and function: Potential implica-
tions for the pathogenesis and prevention of psychopathol-
Damgaard, I.N., Chellakooty, M., . . . Skakkebaek, N.E.
ogy. Prostaglandins Leukotrienes and Essential Fatty Acids,
(2006). Human breast milk contamination with phthalates
75, 329–349.
and alterations of endogenous reproductive hormones in
Megson, D., O’Sullivan, G., Comber, S., Worsfold, P.J., Lohan,
infants three months of age. Environmental Health Perspec-
M.C., Edwards, M.R., . . . Patterson, D.G. (2013). Elucidating
tives, 114, 270–276.
the structural properties that influence the persistence of
Makki, K., Froguel, P., & Wolowczuk, I. (2013). Adipose tissue
PCBs in humans using the national health and nutrition
in obesity-related inflammation and insulin resistance:
examination survey (NHANES) dataset. Science of the Total
Cells, cytokines, and chemokines. ISRN Inflammation,
Environment, 461–462, 99–107.
2013, 139239.
Meyer, U., Yee, B.K., & Feldon, J. (2007). The neurodevelop-
Malm, H., Brown, A.S., Gissler, M., Gyllenberg, D., Hinkka-Yli-
mental impact of prenatal infections at different times of
Salomaki, S., McKeague, I.W., . . . Sourander, A. (2016). Ges-
pregnancy: The earlier the worse? The Neuroscientist: A
tational exposure to selective serotonin reuptake inhibitors
Review Journal Bringing Neurobiology, Neurology and Psy-
and offspring psychiatric disorders: A national register-
chiatry, 13, 241–256.
based study. Journal of the American Academy of Child
Millard, K.N., Frazer, D.M., Wilkins, S.J., & Anderson, G.J.
and Adolescent Psychiatry, 55, 359–366.
(2004). Changes in the expression of intestinal iron trans-
Man, K. K., Tong, H.H., Wong, L.Y., Chan, E.W., Simonoff E.,
port and hepatic regulatory molecules explain the
& Wong, I.C. (2015). Exposure to selective serotonin reup- enhanced iron absorption associated with pregnancy in the
take inhibitors during pregnancy and risk of autism spec- rat. Gut, 53, 655–660.
trum disorder in children: A systematic review and meta- Miodovnik, A., Engel, S.M., Zhu, C.B., Ye, X.Y., Soorya, L.V.,
analysis of observational studies. Neurosci Biobehav Rev 49, Silva, M.J., . . . Wolff, M.S. (2011). Endocrine disruptors and
82–89. childhood social impairment. Neurotoxicology, 32, 261–267.
Mann, J.R., McDermott, S., Bao, H., Hardin, J., & Gregg, A. Miodovnik, A., Edwards, A., Bellinger, D.C., & Hauser, R.
(2010). Pre-eclampsia, birth weight, and autism spectrum (2014). Developmental neurotoxicity of ortho-phthalate
disorders. Journal of Autism and Developmental Disorders, diesters: Review of human and experimental evidence. Neu-
40, 548–554. rotoxicology, 41, 112–122.
Marks, A. R., Harley K., Bradman, A., Kogut, K., Barr D.B., Moore, S. J., Turnpenny, P., Quinn A., Glover, S., Lloyd, D.J.,
Johnson, C., . . . Eskenazi, B. (2010). Organophosphate pesti- Montgomery, T., & J. C. Dean (2000). A clinical study of 57
cide exposure and attention in young Mexican-American children with fetal anticonvulsant syndromes.” J Med
children: The CHAMACOS study. Environ Health Perspect Genet 37, 489–497.
118(12), 1768–1774. Morgenstern, H. (1995). Ecologic studies in epidemiology:
Marsit, C.J. (2016). Placental epigenetics in children’s environ- Concepts, principles, and methods. Annual Review of Pub-
mental health. Seminars in Reproductive Medicine, 34, 36– lic Health, 16, 61–81.
41. MRC Vitamin Study Research Group. (1991). Prevention of
Martin, J.A., Hamilton, B.E., Ventura, S.J., Osterman, M.J., neural tube defects: Results of the medical research council
Kirmeyer, S., Mathews, T.J., . . . Wilson, E.C. (2011). Births: vitamin study. Lancet, 338, 131–137.
Final data for 2009. National Vital Statistics Reports, 60, 1– Nahum Sacks, K., Friger, M., Shoham-Vardi, I., Abokaf, H.,
70. Spiegel, E., Sergienko, R., . . . Sheiner, E. (2016). Prenatal

28 Hertz-Picciotto et al./Environmental contributions to ASD INSAR


Licensed to Mariana Wollinger Maciel Fassheber - marianawmaciel@gmail.com - 046.925.309-60

exposure to gestational diabetes mellitus as an independent and environmental factors in the united states, 1999–2006.
risk factor for long-term neuropsychiatric morbidity of the American Journal of Epidemiology, 181, 171–179.
offspring. American Journal of Obstetrics and Gynecology, Patel, C.J. (2016). Analytical complexity in detection of gene
215, 380 e381–387. variant-by-environment exposure interactions in high-
National Academies of Science EaM. (2015). Mental disorders throughput genomic and exposomic research. Current Envi-
and disabilities among low income children. Washington, ronmental Health Reports, 3, 64–72.
D.C.: National Academies Press. Patel, S., Park, H., Bonato, P., Chan, L., & Rodgers, M. (2012).
National Research Council CotHRoP. (2008). Phthalates and A review of wearable sensors and systems with application
cumulative risk assessment, the tasks ahead. Washington, in rehabilitation. Journal of Neuroengineering and Rehabili-
D.C.: National Academies Press. tation, 9, 21.
Needleman, H.L., Gunnoe, C., Leviton, A., Reed, R., Peresie, Patel, S., Roncaglia, P., & Lovering, R.C. (2015). Using gene
H., Maher, C., . . . Barrett, P. (1979). Deficits in psychologic ontology to describe the role of the neurexin-neuroligin-
and classroom performance of children with elevated den- shank complex in human, mouse and rat and its relevance
tine lead levels. New England Journal of Medicine, 300, to autism. BMC Bioinformatics, 16, 186.
689–695. Patterson, P.H. (2011). Maternal infection and immune
Nelson, K.B. (1991). Prenatal and perinatal factors in the etiol- involvement in autism. Trends in Molecular Medicine, 17,
ogy of autism. Pediatrics, 87, 761–766. 389–394.
O’Brien, K.O., Zavaleta, N., Abrams, S.A., & Caulfield, L.E. Peet, M., Laugharne, J.D., Mellor, J., & Ramchand, C.N. (1996).
(2003). Maternal iron status influences iron transfer to the Essential fatty acid deficiency in erythrocyte membranes
fetus during the third trimester of pregnancy. American from chronic schizophrenic patients, and the clinical effects
Journal of Clinical Nutrition, 77, 924–930. of dietary supplementation. Prostaglandins, Leukotrienes,
O’Rourke, K.M., Redlinger, T.E., & Waller, D.K. (2000). Declin- and Essential Fatty Acids, 55, 71–75.
ing levels of erythrocyte folate during the postpartum Philippat, C., Bennett, D.H., Krakowiak, P., Rose, M., Hwang,
period among hispanic women living on the Texas-Mexico H.-M., & Hertz-Picciotto, I. (2015). Phthalate concentrations
border. Journal of Women’s Health & Gender-Based Medi- in house dust in relation to autism spectrum disorder and
cine, 9, 397–403. developmental delay in the childhood autism risks from
Oken, E., Radesky, J.S., Wright, R.O., Bellinger, D.C., genetics and the environment (charge) study. Environmen-
Amarasiriwardena, C.J., Kleinman, K.P., . . . Gillman, M.W. tal Health, 14, 56.
(2008). Maternal fish intake during pregnancy, blood mer- Pickles, A., Bolton, P., Macdonald, H., Bailey, A., Le Couteur,
cury levels, and child cognition at age 3 years in a US A., Sim, C.H., . . . Rutter, M. (1995). Latent-class analysis of
cohort. American Journal of Epidemiology, 167, 1171–1181. recurrence risks for complex phenotypes with selection and
Onore, C., Careaga, M., & Ashwood, P. (2012). The role of measurement error: A twin and family history study of
immune dysfunction in the pathophysiology of autism. autism. American Journal of Human Genetics, 57, 717–726.
Brain Behavior and Immunity, 26, 383–392. Piven, J., Berthier, M.L., Starkstein, S.E., Nehme, E., Pearlson,
Osterman, M.J., & Martin, J.A. (2014). Trends in low-risk cesar- G., & Folstein, S. (1990). Magnetic resonance imaging evi-
ean delivery in the united states, 1990–2013. National Vital dence for a defect of cerebral cortical development in
Statistics Reports, 63, 1–16. autism. American Journal of Psychiatry, 147, 734–739.
Pantham, P., Aye, I.L., & Powell, T.L. (2015). Inflammation in Pollitt, E. (1993). Iron deficiency and cognitive function.
maternal obesity and gestational diabetes mellitus. Pla- Annual Review of Nutrition, 13, 521–537.
centa, 36, 709–715. Post, G.B., Louis, J.B., Lippincott, R.L., & Procopio, N.A.
Parracho, H.M., Bingham, M.O., Gibson, G.R., & McCartney, (2013). Occurrence of perfluorinated compounds in raw
A.L. (2005). Differences between the gut microflora of chil- water from new jersey public drinking water systems. Envi-
dren with autistic spectrum disorders and that of healthy ronmental Science and Technology, 47, 13266–13275.
children. Journal of Medical Microbiology, 54, 987–991. Rai, D., Lee, B.K., Dalman, C., Golding, J., Lewis, G., &
Patel, C.J., Bhattacharya, J., & Butte, A.J. (2010). An Magnusson, C. (2013). Parental depression, maternal anti-
environment-wide association study (EWAS) on type 2 dia- depressant use during pregnancy, and risk of autism spec-
betes mellitus. PLoS One, 5, e10746. trum disorders: Population based case-control study. BMJ,
Patel, C.J., Chen, R., & Butte, A.J. (2012). Data-driven integra- 346, f2059.
tion of epidemiological and toxicological data to select can- Rao, T.J., & Province, M.A. (2016). A framework for interpret-
didate interacting genes and environmental factors in ing type i error rates from a product-term model of interac-
association with disease. Bioinformatics, 28, i121–i126. tion applied to quantitative traits. Genetic Epidemiology,
Patel, C.J., Chen, R., Kodama, K., Ioannidis, J.P., & Butte, A.J. 40, 144–153.
(2013). Systematic identification of interaction effects Rauh, V. A., Garfinkel, R., Perera, F.P., Andrews, H.F., Hoepner,
between validated genome- and environment-wide associa- L., Barr, D.B., . . . Whyatt R.W. (2006). Impact of prenatal
tions on type 2 diabetes mellitus. AMIA Joint Summits on chlorpyrifos exposure on neurodevelopment in the first 3
Translational Science Proceedings AMIA Joint Summits on years of life among inner-city children. Pediatrics 118(6),
Translational Science, 2013, 135. e1845–1859.
Patel, C.J., Ioannidis, J.P., Cullen, M.R., & Rehkopf, D.H. Rauh, V., Arunajadai, S., Horton, M., Perera, F., Hoepner, L.,
(2015). Systematic assessment of the correlations of house- Barr, D.B., . . . Whyatt, R. (2011). Seven-year neurodevelop-
hold income with infectious, biochemical, physiological, mental scores and prenatal exposure to chlorpyrifos, a

INSAR Hertz-Picciotto et al./Environmental contributions to ASD 29


Licensed to Mariana Wollinger Maciel Fassheber - marianawmaciel@gmail.com - 046.925.309-60

common agricultural pesticide. Environmental Health Per- analytes from microneedle patches. Analytical Chemistry,
spectives, 119, 1196–1201. 86, 10520–10523.
Rauh, V.A., Perera, F.P., Horton, M.K., Whyatt, R.M., Bansal, Rombaldi Bernardi, J., de Souza Escobar, R., Ferreira, C.F., &
R., Hao, X., . . . Peterson, B.S. (2012). Brain anomalies in Pelufo Silveira, P. (2012). Fetal and neonatal levels of
children exposed prenatally to a common organophosphate omega-3: Effects on neurodevelopment, nutrition, and
pesticide. Proceedings of the National Academy of Sciences growth. The Scientific World Journal, 2012, 202473.
of the United States of America, 109, 7871–7876. Rosen, B.N., Lee, B.K., Lee, N.L., Yang, Y., & Burstyn, I. (2015).
Raz, R., Roberts, A.L., Lyall, K., Hart, J.E., Just, A.C., Laden, F., Maternal smoking and autism spectrum disorder: A meta-
. . . Weisskopf, M.G. (2015). Autism spectrum disorder and analysis. Journal of Autism and Developmental Disorders,
particulate matter air pollution before, during, and after 45, 1689–1698.
pregnancy: A nested case-control analysis within the nurses’ Roth, C., Magnus, P., Schjolberg, S., Stoltenberg, C., Suren, P.,
health study ii cohort. Environmental Health Perspectives, McKeague, I.W., . . . Susser, E. (2011). Folic acid supple-
123, 264–270. ments in pregnancy and severe language delay in children.
Reik, W., & Walter, J. (2001). Genomic imprinting: Parental JAMA, 306, 1566–1573.
influence on the genome. Nature Reviews Genetics, 2, 21– Rothman, K.J., Greenland, S., & Lash, T.L. 2008. Modern epi-
32. demiology. 3rd ed. Philadelphia, PA: Lippincott Williams &
Reiner, O., Karzbrun, E., Kshirsagar, A., & Kaibuchi, K. (2016). Wilkins.
Regulation of neuronal migration, an emerging topic in Roza, S.J., van Batenburg-Eddes, T., Steegers, E.A., Jaddoe,
autism spectrum disorders. Journal of Neurochemistry, 136, V.W., Mackenbach, J.P., Hofman, A., . . . Teimeier, H.
440–456. (2010). Maternal folic acid supplement use in early preg-
Richardson, A.J., Calvin, C.M., Clisby, C., Schoenheimer, D.R., nancy and child behavioral problems: The generation R
Montgomery, P., Hall, J.A., . . . Stein, J.F. (2000a). Fatty acid study. British Journal of Nutrition, 103, 445–452.
deficiency signs predict the severity of reading and related Sagiv, S.K., Thurston, S.W., Bellinger, D.C., Amarasiriwardena,
difficulties in dyslexic children. Prostaglandins, Leuko- C., & Korrick, S.A. (2012). Prenatal exposure to mercury
trienes, and Essential Fatty Acids, 63, 69–74. and fish consumption during pregnancy and attention-defi-
Richardson, A.J., Easton, T., & Puri, B.K. (2000b). Red cell and cit/hyperactivity disorder-related behavior in children.
plasma fatty acid changes accompanying symptom remis- Archives of Pediatrics & Adolescent Medicine, 166, 1123–
sion in a patient with schizophrenia treated with eicosapen- 1131.
taenoic acid. European Neuropsychopharmacology, 10, Sagiv, S.K., Kogut, K., Gaspar, F.W., Gunier, R.B., Harley, K.G.,
189–193. Parra, K., . . . Eskenazi, B. (2015). Prenatal and childhood
Richardson, A.J., & Ross, M.A. (2000). Fatty acid metabolism in polybrominated diphenyl ether (PBDE) exposure and atten-
neurodevelopmental disorder: A new perspective on associ- tion and executive function at 9–12years of age. Neurotoxi-
ations between attention-deficit/hyperactivity disorder, dys- cology Teratology, 52, 151–161.
lexia, dyspraxia and the autistic spectrum. Prostaglandins Schaevitz, L.R., & Berger-Sweeney, J.E. (2012). Gene-environ-
Leukotrienes and Essential Fatty Acids, 63, 1–9. ment interactions and epigenetic pathways in autism: The
Rimland, B. 1964. Infantile autism: The syndrome and its importance of one-carbon metabolism. ILAR Journal, 53,
implications for a neural theory of autism. New York: 322–340.
Appleton-Century-Croft. Schafer, K.S., & Kegley, S.E. (2002). Persistent toxic chemicals
Ritvo, E.R., Jorde, L.B., Mason-Brothers, A., Freeman, B.J., in the US food supply. Journal of Epidemiology and Com-
Pingree, C., Jones, M.B., . . . Mo, A. (1989). The UCLA- munity Health, 56, 813–817.
university of Utah epidemiologic survey of autism: Recur- Schlotz, W., Jones, A., Phillips, D.I., Gale, C.R., Robinson,
rence risk estimates and genetic counseling. American Jour- S.M., & Godfrey, K.M. (2010). Lower maternal folate status
nal of Psychiatry, 146, 1032–1036. in early pregnancy is associated with childhood hyperactiv-
Roberts, E.M., English, P.B., Grether, J.K., Windham, G.C., ity and peer problems in offspring. Journal of Child Psy-
Somberg, L., & Wolff, C. (2007). Maternal residence near chology and Psychiatry, 51, 594–602.
agricultural pesticide applications and autism spectrum dis- Schmidt, R.J., Hansen, R.L., Hartiala, J., Allayee, H., Schmidt,
orders among children in the California central valley. L.C., Tancredi, D.J., . . . Hertz-Picciotto, I. (2011). Prenatal
Environmental Health Perspectives, 115, 1482–1489. vitamins, one-carbon metabolism gene variants, and risk
Roberts, A. L., Lyall, K., Hart, J.E., Laden, F., Just, A.C., Bobb, for autism. Epidemiology, 22, 476–485.
J.F., . . . Weisskopf M.G. (2013). Perinatal air pollutant expo- Schmidt, R.J., Tancredi, D.J., Ozonoff, S., Hansen, R.L.,
sures and autism spectrum disorder in the children of Hartiala, J., Allayee, H., . . . Hertz-Picciotto, I. (2012). Mater-
nurses’ health study II participants. Environ Health Perspect nal periconceptional folic acid intake and risk of autism
121, 978–984. spectrum disorders and developmental delay in the charge
Rodier, P.M., Ingram, J.L., Tisdale, B., Nelson, S., & Romano, J. (childhood autism risks from genetics and environment)
(1996). Embryological origin for autism: Developmental case-control study. American Journal of Clinical Nutrition,
anomalies of the cranial nerve motor nuclei. Journal of 96, 80–89.
Comparative Neurology, 370, 247–261. Schmidt, R.J., Lyall, K., & Hertz-Picciotto, I. (2014a). Environ-
Romanyuk, A.V., Zvezdin, V.N., Samant, P., Grenader, M.I., ment and autism: Current state of the science. Cutting
Zemlyanova, M., & Prausnitz, M.R. (2014). Collection of Edge Psychiatry Practice, 4, 21–38.

30 Hertz-Picciotto et al./Environmental contributions to ASD INSAR


Licensed to Mariana Wollinger Maciel Fassheber - marianawmaciel@gmail.com - 046.925.309-60

Schmidt, R.J., Tancredi, D.J., Krakowiak, P., Hansen, R.L., & Steenweg-de Graaff, J., Ghassabian, A., Jaddoe, V.W., Tiemeier,
Ozonoff, S. (2014b). Maternal intake of supplemental iron H., & Roza, S.J. (2015). Folate concentrations during preg-
and risk of autism spectrum disorder. American Journal of nancy and autistic traits in the offspring. The generation R
Epidemiology, 180, 890–900. study. European Journal of Public Health, 25, 431–433.
Schmidt, R.J., Schroeder, D.I., Crary-Dooley, F.K., Barkoski, Steenweg-de Graaff, J., Tiemeier, H., Ghassabian, A.,
J.M., Tancredi, D.J., Walker, C.K., . . . LaSalle, J.M. (2016). Rijlaarsdam, J., Jaddoe, V.W., Verhulst, F.C., . . . Roza, S.J.
Self-reported pregnancy exposures and placental DNA (2016). Maternal fatty acid status during pregnancy and
methylation in the marbles prospective autism sibling child autistic traits: The generation r study. American Jour-
study. Environmental Epigenetics, 2, 1–10. nal of Epidemiology, 183, 792–799.
Schmidt, R.J., Kogan, V., Shelton, J.F., Delwiche, L., Hansen, Stein, C.R., & Savitz, D.A. (2011). Serum perfluorinated com-
R.L., Ozonoff, S., . . . LaSalle, J.M. (2017). Combined prena- pound concentration and attention deficit/hyperactivity
tal pesticide exposure and folic acid intake in relation to disorder in children 5–18 years of age. Environmental
autism spectrum disorder. Environmental Health Perspec- Health Perspectives, 119, 1466–1471.
tives, 125, 097007. Stern, M. (2011). Insulin signaling and autism. Front Endocri-
Schroeder, D.I., Lott, P., Korf, I., & LaSalle, J.M. (2011). Large- nol (Lausanne), 2, 54.
scale methylation domains mark a functional subset of neu- Stoch, Y.K., Williams, C.J., Granich, J., Hunt, A.M., Landau,
ronally expressed genes. Genome Research, 21, 1583–1591. L.I., Newnham, J.P., . . . Whitehouse, A.J.. (2012). Are prena-
Schroeder, D.I., Schmidt, R.J., Crary-Dooley, F.K., Walker, C.K., tal ultrasound scans associated with the autism phenotype?
Ozonoff, S., Tancredi, D.J., . . . LaSalle, J.M. (2016). Placental Follow-up of a randomised controlled trial. Journal of
methylome analysis from a prospective autism study. Autism and Developmental Disorders, 42, 2693–2701.
Molecular Autism, 7, 51. Stoltzfus, R. (2001). Defining iron-deficiency anemia in public
Schuh, R.A., Lein, P.J., Beckles, R.A., & Jett, D.A. (2002). Non- health terms: A time for reflection. Journal of Nutrition,
131, 565S–567S.
cholinesterase mechanisms of chlorpyrifos neurotoxicity:
Strain, J.J., Davidson, P.W., Bonham, M.P., Duffy, E.M., Stokes-
Altered phosphorylation of ca21/camp response element
Riner, A., Thurston, S.W., . . . Clarkson, T.W. (2008). Associ-
binding protein in cultured neurons. Toxicology and
ations of maternal long-chain polyunsaturated fatty acids,
Applied Pharmacology, 182, 176–185.
methyl mercury, and infant development in the Seychelles
Shelton, J.F., Tancredi, D.J., & Hertz-Picciotto, I. (2010). Inde-
child development nutrition study. Neurotoxicology, 29,
pendent and dependent contributions of advanced mater-
776–782.
nal and paternal ages to autism risk. Autism Research, 3,
Stromland, K., Nordin, V., Miller, M., Akerstrom, B., &
30–39.
Gillberg, C. (1994). Autism in thalidomide embryopathy: A
Shelton, J.F., Hertz-Picciotto, I., & Pessah, I.N. (2012). Tipping
population study. Developmental Medicine and Child Neu-
the balance of autism risk: Potential mechanisms linking
rology, 36, 351–356.
pesticides and autism. Environmental Health Perspectives,
Suren, P., Roth, C., Bresnahan, M., Haugen, M., Hornig, M.,
120, 944–951.
Hirtz, D., . . . Stoltenberg, C. (2013). Association between
Shelton, J.F., Geraghty, E.M., Tancredi, D.J., Delwiche, L.D.,
maternal use of folic acid supplements and risk of autism
Schmidt, R.J., Ritz, B., . . . Picciotto, I.H. (2014). Neurodeve-
spectrum disorders in children. JAMA, 309, 570–577.
lopmental disorders and prenatal residential proximity to
Swamy, G.K., Reddick, K.L., Brouwer, R.J., Pollak, K.I., &
agricultural pesticides: The charge study. Environmental Myers, E.R. (2011). Smoking prevalence in early pregnancy:
Health Perspectives, 122, 1103–1109. Comparison of self-report and anonymous urine cotinine
Shipton, D., Tappin, D.M., Vadiveloo, T., Crossley, J.A., Aitken, testing. The Journal of Maternal-Fetal & Neonatal Medicine,
D.A., & Chalmers, J. (2009). Reliability of self-reported 24, 86–90.
smoking status by pregnant women for estimating smoking Talbott, E.O., Marshall, L.P., Rager, J.R., Arena, V.C., Sharma,
prevalence: A retrospective, cross sectional study. BMJ, 339, R.K., & Stacy, S.L. (2015). Air toxics and the risk of autism
b4347. spectrum disorder: The results of a population based case-
Smalley, S.L., Tanguay, P.E., Smith, M., & Gutierrez, G. (1992). control study in southwestern Pennsylvania. Environmental
Autism and tuberous sclerosis. Journal of Autism Develop- Health, 14, 80.
mental Disorder, 22, 339–355. Tang, S., Wang, Y., Gong, X., & Wang, G. (2015). A meta-
Smits, L.J., & Essed, G.G. (2001). Short interpregnancy inter- analysis of maternal smoking during pregnancy and autism
vals and unfavourable pregnancy outcome: Role of folate spectrum disorder risk in offspring. International Journal of
depletion. Lancet, 358, 2074–2077. Environmental Research and Public Health, 12, 10418–
Starling, P., Charlton, K., McMahon, A.T., & Lucas, C. (2015). 10431.
Fish intake during pregnancy and foetal neurodevelop- Tau, G.Z., & Peterson, B.S. (2010). Normal development of
ment–a systematic review of the evidence. Nutrients, 7, brain circuits. Neuropsychopharmacology, 35, 147–168.
2001–2014. Taylor, K.W., Joubert, B.R., Braun, J.M., Dilworth, C.,
Steegers-Theunissen, R.P., Obermann-Borst, S.A., Kremer, D., Gennings, C., Hauser, R., . . . Carlin, D.J. (2016). Statistical
Lindemans, J., Siebel, C., Steegers, E.A., . . . Heijmans, B.T. approaches for assessing health effects of environmental
(2009). Periconceptional maternal folic acid use of 400 chemical mixtures in epidemiology: Lessons from an inno-
microg per day is related to increased methylation of the vative workshop. Environmental Health Perspectives, 124,
IGF2 gene in the very young child. PLoS One, 4, e7845. A227–A229.

INSAR Hertz-Picciotto et al./Environmental contributions to ASD 31


Licensed to Mariana Wollinger Maciel Fassheber - marianawmaciel@gmail.com - 046.925.309-60

Tchernia, G., Archambeaud, M.P., Yvart, J., & Diallo, D. Wang, Z., Hong, Y., Zou, L., Zhong, R., Zhu, B., Shen, N., . . .
(1996). Erythrocyte ferritin in human neonates: Maternofe- Miao, X. (2014). Reelin gene variants and risk of autism
tal iron kinetics revisited. Clinical and Laboratory Haema- spectrum disorders: An integrated meta-analysis. American
tology, 18, 147–153. Journal of Medical Genetics Part B Neuropsychiatric Genet-
U.S. Environmental Protection Agency. (2017). EPA administra- ics, 165B, 192–200.
tor Pruitt Denies petition to ban widely used pesticide. Wassink, T.H., & Piven, J. (2000). The molecular genetics of
Washington, DC: U.S. EPA Media Relations. autism. Current Psychiatry Reports, 2, 170–175.
U.S. Environmental Protection Agency. Office of Prevention Wassink, T.H., Piven, J., & Patil, S.R. (2001). Chromosomal
PaTS. (2006). Action memorandum. Inert reassessments: abnormalities in a clinic sample of individuals with autistic
One exemption from the requirement of a tolerance for disorder. Psychiatric Genetics, 11, 57–63.
diethyl phthalate. Washington, DC: U.S. Environmental Wassink, T.H., Brzustowicz, L.M., Bartlett, C.W., & Szatmari, P.
Protection Agency. (2004). The search for autism disease genes. Mental Retarda-
von Ehrenstein, O.S., Aralis, H., Cockburn, M., & Ritz, B. tion and Developmental Disabilities Research Reviews, 10,
(2014). In utero exposure to toxic air pollutants and risk of 272–283.
childhood autism. Epidemiology 25(6), 851–858. Webb, S.J., Garrison, M.M., Bernier, R., McClintic, A.M., King,
van Eijsden, M., Smits, L.J., van der Wal, M.F., & Bonsel, G.J. B.H., & Mourad, P.D. (2017). Severity of ASD symptoms
(2008). Association between short interpregnancy intervals and their correlation with the presence of copy number var-
and term birth weight: The role of folate depletion. Ameri- iations and exposure to first trimester ultrasound. Autism
can Journal of Clinical Nutrition, 88, 147–153. Research, 10, 472–484.
Vargas, D.L., Nascimbene, C., Krishnan, C., Zimmerman, A.W., Whyatt, R.M., Liu, X., Rauh, V.A., Calafat, A.M., Just, A.C.,
& Pardo, C.A. (2005). Neuroglial activation and neuroin- Hoepner, L., . . . Factor-Litvak, X. (2012). Maternal prenatal
flammation in the brain of patients with autism. Annals of urinary phthalate metabolite concentrations and child
Neurology, 57, 67–81. mental, psychomotor, and behavioral development at 3
Virk, J., Liew, Z., Olsen, J., Nohr, E.A., Catov, J.M., & Ritz, B. years of age. Environmental Health Perspectives, 120, 290–
(2016). Preconceptional and prenatal supplementary folic 295.
acid and multivitamin intake and autism spectrum disor- Wiest, M.M., German, J.B., Harvey, D.J., Watkins, S.M., &
ders. Autism, 20, 710–718. Hertz-Picciotto, I. (2009). Plasma fatty acid profiles in
Vogel Ciernia, A., & LaSalle, J. (2016). The landscape of DNA autism: A case-control study. Prostaglandins Leukotrienes
methylation amid a perfect storm of autism aetiologies. Essential Fatty Acids, 80, 221–227.
Nature Reviews Neuroscience, 17, 411–423. Williams, E.L., & Casanova, M.F. (2010). Potential teratogenic
Vogt, R., Bennett, D., Cassady, D., Frost, J., Ritz, B., & Hertz- effects of ultrasound on corticogenesis: Implications for
Picciotto, I. (2012). Cancer and non-cancer health effects autism. Medical Hypotheses, 75, 53–58.
from food contaminant exposures for children and adults Windham, G., Zhang, L., Gunier, R., Croen, L., & Grether, J.
in california: A risk assessment. Environmental Health, 11, (2006). Autism spectrum disorders in relation to distribu-
83. tion of hazardous air pollutants in the San Francisco bay
Volk, H.E., Hertz-Picciotto, I., Delwiche, L., Lurmann, F., & area. Environmental Health Perspectives, 114, 1438–1444.
McConnell, R. (2011). Residential proximity to freeways Winneke, G., Bucholski, A., Heinzow, B., Kramer, U., Schmidt,
and autism in the charge study. Environmental Health Per- € ber H.J. (1998). Developmen-
E., Walkowiak, J., . . . Steingru
spectives, 119, 873–877. tal neurotoxicity of polychlorinated biphenyls (PCBs): Cog-
Volk, H.E., Lurmann, F., Penfold, B., Hertz-Picciotto, I., & nitive and psychomotor functions in 7-month old children.
McConnell, R. (2012). Traffic related air pollution, particu- Toxicology Letters, 102–103, 423–428.
late matter, and autism. JAMA Psychiatry, 70, 71–77. Witte, J.S., Greenland, S., Haile, R.W., & Bird, C.L. (1994).
Volk, H.E., Kerin, T., Lurmann, F., Hertz-Picciotto, I., Hierarchical regression analysis applied to a study of multi-
McConnell, R., & Campbell, D.B. (2014). Autism spectrum ple dietary exposures and breast cancer. Epidemiology, 5,
disorder: Interaction of air pollution with the met receptor 612–621.
tyrosine kinase gene. Epidemiology, 25, 44–47. Wofford, P., Segawa, R., Schreider, J., Federighi, V., Neal, R., &
Voorhees, J.R., Rohlman, D.S., Lein, P.J., & Pieper, A.A. (2016). Brattesani, M. (2014). Community air monitoring for pesti-
Neurotoxicity in preclinical models of occupational expo- cides. Part 3: Using health-based screening levels to evalu-
sure to organophosphorus compounds. Frontiers in Neuro- ate results collected for a year. Environmental Monitoring
science, 10, 590. Assessment, 186, 1355–1370.
Vreugdenhil, H.J., Lanting, C.I., Mulder, P.G., Boersma, E.R., & Wolff, G.L., Kodell, R.L., Moore, S.R., & Cooney, C.A. (1998).
Weisglas-Kuperus, N. (2002). Effects of prenatal PCB and Maternal epigenetics and methyl supplements affect
dioxin background exposure on cognitive and motor abili- agouti gene expression in Avy/a mice. Faseb Journal, 12,
ties in Dutch children at school age. Journal of Pediatrics, 949–957.
140, 48–56. Wong, C.T., Ahmad, E., Li, H., & Crawford, D.A. (2014). Pros-
Walker, C.K., Krakowiak, P., Baker, A., Hansen, R.L., Ozonoff, taglandin E2 alters Wnt-dependent migration and prolifera-
S., & Hertz-Picciotto, I. (2015). Preeclampsia, placental tion in neuroectodermal stem cells: Implications for autism
insufficiency, and autism spectrum disorder or developmen- spectrum disorders. Cell Communication and Signaling, 12,
tal delay. JAMA Pediatrics, 169, 154–162. 19.

32 Hertz-Picciotto et al./Environmental contributions to ASD INSAR


Licensed to Mariana Wollinger Maciel Fassheber - marianawmaciel@gmail.com - 046.925.309-60

Wu, X.M., Bennett, D.H., Moran, R.E., Sjodin, A., Jones, R.S., delays? Results from the charge (childhood autism risks
Tancredi, D.J., . . . Picciotto I.H. (2015). Polybrominated from genetics and environment) study. Journal of Autism
diphenyl ether serum concentrations in a Californian popu- Developmental Disorders, 43, 25–33.
lation of children, their parents, and older adults: An expo- Zerbo, O., Yoshida, C., Gunderson, E.P., Dorward, K., & Croen,
sure assessment study. Environmental Health, 14, 23. L.A. (2015). Interpregnancy interval and risk of autism spec-
Xiang, A.H., Wang, X., Martinez, M.P., Walthall, J.C., Curry, trum disorders. Pediatrics, 136, 651–657.
E.S., Page, K., . . . Getahun D. (2015). Association of mater- Zhang, Z., Cao, M., Chang, C.W., Wang, C., Shi, X., Zhan, X.,
nal diabetes with autism in offspring. JAMA, 313, 1425– . . . Wu, J.I. (2016). Autism-associated chromatin regulator
1434. brg1/smarca4 is required for synapse development and
Xu, G., Jing, J., Bowers, K., Liu, B., & Bao, W. (2014). Maternal myocyte enhancer factor 2-mediated synapse remodeling.
diabetes and the risk of autism spectrum disorders in the Molecular and Cellular Biology, 36, 70–83.
offspring: A systematic review and meta-analysis. Journal of Zhao, X., Wang, H., Li, J., Shan, Z., Teng, W., & Teng, X.
Autism Developmental Disorders, 44, 766–775. (2015). The correlation between polybrominated diphenyl
Yao, Y., Walsh, W.J., McGinnis, W.R., & Pratico, D. (2006). ethers (PBDEs) and thyroid hormones in the general popu-
Altered vascular phenotype in autism: Correlation with oxi- lation: A meta-analysis. PLoS One, 10, e0126989.
dative stress. Archives of Neurology, 63, 1161–1164. Zheng, Y.L., Ding, X.R., Poon, C.C., Lo, B.P., Zhang, H., Zhou,
Yui, K., Imataka, G., Kawasaki, Y., & Yamada, H. (2016). X.L., . . . Zhang, Y.T. (2014). Unobtrusive sensing and wear-
Down-regulation of a signaling mediator in association able devices for health informatics. IEEE Transactions on
with lowered plasma arachidonic acid levels in individuals Bio-Medical Engineering, 61, 1538–1554.
with autism spectrum disorders. Neuroscience Letters, 610, Zhong, J., Karlsson, O., Wang, G., Li, J., Guo, Y., Lin, X., . . .
223–228. Baccarelli, A.A. (2017). B vitamins attenuate the epigenetic
Zeidan-Chulia, F., de Oliveira, B.H., Salmina, A.B., Casanova, effects of ambient fine particles in a pilot human interven-
M.F., Gelain, D.P., Noda, M., . . . Moreira, J.C. (2014). Altered tion trial. Proceedings of the National Academy of Sciences
expression of Alzheimer’s disease-related genes in the cere- of the United States of America, 114, 3503–3508.
bellum of autistic patients: A model for disrupted brain con- Zhu, Z., Liu, T., Li, G., Li, T., & Inoue, Y. (2015). Wearable sen-
nectome and therapy. Cell Death & Disease, 5, e1250. sor systems for infants. Sensors, 15, 3721–3749.
Zerbo, O., Iosif, A.M., Delwiche, L., Walker, C., & Hertz- Zhubi, A., Cook, E.H., Guidotti, A., & Grayson, D.R. (2014).
Picciotto, I. (2011). Month of conception and risk of Epigenetic mechanisms in autism spectrum disorder. Inter-
autism. Epidemiology, 22, 469–475. national Review of Neurobiology, 115, 203–244.
Zerbo, O., Iosif, A.M., Walker, C., Ozonoff, S., Hansen, R.L., & Zou, H., & Hastie, T. (2005). Regularization and variable selec-
Hertz-Picciotto, I. (2012). Is maternal influenza or fever dur- tion via the elastic net. Journal of the Royal Statistical Soci-
ing pregnancy associated with autism or developmental ety Series B, 67, 301–320.

You might also like