The importance of research on

environmental factors in a highly

heritable disorder

Mark A. Corrales, MPP

US Environmental Protection Agency
Statements are those of the author, and
do not necessarily represent views or policies of the EPA

Autism One Conference

May 22, 2009
 Parallel Universes?
• General public
• Parents & affected kids
• Interest groups:
– environmental health advocacy
– industries
• Researchers
• Policy/ decision-makers:
Lawyers, economists, policy and risk
analysts, staff scientists, etc.
 Getting from here to there
• If you believe national-level
medical and environmental health
interventions are needed…
• the autism research literature
is a key bottleneck.
 From evidence to action
Drug
development

RCTs Behav./Educ.
Experimental
therapy
toxicology,
guidelines
mechanisms
Prospective
epidemiology Regulations
Retrospective to limit risk
epidemiology factors
Ecologic
Anecdotes, studies
More studies
case studies
Better studies
Guesses,
theories More types of studies
(multiple lines of evidence)

Public & scientific consensus

 Is there sufficient evidence
for action?

Published scientific research

literature on environmental factors
in autism is still quite limited
compared to other diseases or
health effects.
 Why so little research on
environmental factors in autism?
• All autism research limited
(until recently)
• Behavioral > molecular level
(until recently)
• Genes > Environment:
High heritability
misinterpreted as
“genes not environment”
• Toxicology and epidemiology not sufficiently
integrated with autism research so far
 Why so little research on
environmental factors in autism?
• All autism research limited
(until recently)
• Behavioral > molecular level
(until recently)
• Genes > Environment:
High heritability
misinterpreted as
“genes not environment”
• Toxicology and epidemiology not sufficiently
integrated with autism research so far
 Most autism science is
extremely recent
As of early 2009,
• Half of all the autism research literature
indexed in PubMed was published
in the past eight years: 2000-2008.
• One third was published in
just the last five years: 2004-2008.
 Very Recent
Explosion of Interest in Autism
In 2008:
• Newspaper articles:
>3,000 headlines/ year
• Scientific journal articles:
>1,000 / year
• New books:
>150 / year
Journal articles
 But so far…
• Treatment --
New ASD treatments approved: 0
• Primary prevention (at national level) --
Environmental standards, regulations,
or other actions taken based on any
potential ASD risk factors: 0
 Autism Literature is Still
Relatively Limited
• If you believe national-level
medical and environmental health
interventions are needed…
• the autism research literature
is a key bottleneck.
What has autism literature
focused on?
 Why so little research on
environmental factors in autism?
• All autism research limited
(until recently)
• Behavioral > molecular level
(until recently)
• Genes > Environment:
High heritability
misinterpreted as
“genes not environment”
• Toxicology and epidemiology not sufficiently
integrated with autism research so far
 Autism literature has focused on
behavioral, not chemical level
• The PDD literature focused on the
behavioral and cognitive levels three
times as much as did biomedical
literature as a whole (as % of articles).
• 56% vs. only 38% of publications were
indexed with chemical substances in
PubMed vs. PDD literature, even
recently (2000-2006).
Corrales, Ringer, & Herbert 2008 IMFAR poster #2999
Moldin vs. Lathe
biological level
 Focus on Chemicals & Drugs Level
% of publications 1985-2007 with

Chem level

100%

75%
MeSH

48%
50%

30%

25%

0%
n

CP
ia
r

IV
TB

]
n

on

D
y
la

H
re

ps
io

H
en

[M
si
po
Pr

AD
Pe

ile

es
hr

D
Bi

Ep

op

r
tis

PD
ep
i

iz
in

D
h
Rh

Sc

Comorbid disorders at elevated prevalence in PDD according to Rzhetsky (2007)

 Why so little research on
environmental factors in autism?
• All autism research limited
(until recently)
• Behavioral > molecular level
(until recently)
• Genetic > Environmental studies:
High heritability
misinterpreted as
“genes not environment”
• Toxicology and epidemiology not sufficiently
integrated with autism research so far
Genetic etiology
pie
 A few molecular substances have
dominated the literature (1985-2007)
• Approx. 2,000 unique substance terms in ASD
literature, occurring 12k times in 4,000 PMIDs
– Half the substance terms appear only once.
– Avg. substance is in just 6 PMIDs
• Chemicals in >100 articles each:
– MECP2 gene or protein (>400)
– Serotonin (251)
– Risperidone (112)

Corrales, Ringer, & Herbert 2008 IMFAR poster #2999.

 Medications vs.
environmental pollutants
• >130 medications in >1200 PMIDs ( ~10% of articles)
• 43 environmental pollutant terms (other than Hg)
occur 111 times across 84 PMIDs ( <1% of articles)
– Only 36/111 refer to a specific chemical rather than class
(“teratogens” or “environmental pollutants” seen up to 13
times each)
– Lead and cocaine were the most common specific toxins.
– No other specific toxin (except Hg) appeared in more than 4
PMIDs. 3 Hg-related terms in 91 PMIDs (<1%)
• 41 dietary substances in 189 PMIDs (1-2%)
(e.g., gluten, folic acid) (excluding food additives)
• A few potential treatments/ causes
(valproic acid, oxytocin, melatonin, secretin) found in
26-91 PMIDs each.
• ~20 other terms in 2-14 PMIDs each:
(food additives, illegal drugs, infectious agents or
allergens, nicotine, alcohol, caffeine)
Corrales, Ringer, & Herbert 2008 IMFAR poster #2999. Includes publications 1985-2007.
 Environment-relevant topics
understudied in ASD literature,
relative to all biomedical literature
 Hazardous Substances + Organophosph.
Haz subst

Chlorpyrifos
Phosphoric Acid Esters
Street Drugs
Cadmium
Asbestos
Pesticides
Water Pollutants, Chemical
Air Pollutants
Mutagens
Insecticides
Cyclophosphamide
Organophosphorus Compounds
Endotoxins
Carcinogens

0.01 0.10 1.00 10.00 100.00

Rate Ratio (Rate in ASD vs all PubMed citations)
 Cell types
Cell Types

Neurons

T-Lymphocytes

Killer Cells, Natural

B-Lymphocytes

Macrophages

0.01 0.10 1.00 10.00 100.00

Rate Ratio (Rate in ASD vs all PubMed citations)
 Understudied in ASD vs.
in 10 comorbid disorders
(1985-2007)

• IL-6: Studied 9x more often in comorbid

literature than in ASD
• Glucocorticoids: 52x
• Anti-Inflammatory Agents: 36x
• Adrenal Cortex Hormones: 17x
• Thyroxine: 7x
• Steroids: 7x
• Herpesviridae Infections: 16x
Corrales, Ringer, & Herbert 2008 IMFAR poster #2999
 Why so little research on
environmental factors in autism?
 Highly-cited statements on
autism and environment
• “Among cognitive diseases,
ASDs are the most heritable
(about 80%), suggesting that
they are determined largely by genes
and not by the environment.”
[emphasis added]
Südhof (2008) Nature
• Heritability is at least 90%.
Moldin, Rubenstein, and Hyman (2006) Journal of
Neuroscience
Genes and Environment
in Autism:
From Pie to Venn
 G AND E
DO NOT FORM A PIE CHART

GENES

ENVIRONMENT

If one is trying to show population attributable fractions (PAF),

the sum is not 100%. If one is trying to partition variance, GxE
should be explained.
 G AND E
FORM A VENN DIAGRAM

G&E
G E
 G AND E
G&E
not G vs. E G E

• “What % is genetic?”
– Variance studies misinterpreted as ruling out E,
or as if heritability were PAF
– Sum of PAFs is not 100%
– A case is not G or E
• G and E are needed to explain the risk
 Tuberculosis as
“a genetic disease”?
• TB was thought/ known to be “a genetic
disease,” because it was seen to run in
families.
• Later, we realized some people are
genetically more susceptible to the
environmental cause (infectious agent).
• TB in the population is caused by the
combination of G and E (“GxE”).
• Many examples of GxE have been discovered
(mostly in cancers & cardiovascular disease,
but also PKU, depression, ADHD, etc.)
 How can environment be
important if heritability is so high?
• GxE (GE interaction)
Most heritability studies assume no GxE.

• epiG E (epigenetic change)

• G E (genetic damage)

• G E (GE correlation)
 I. Gene-environment
interaction (GxE)
• TB example
• Most heritability estimates ignore GxE Shared

• Heritability is overestimated where genes

interact with shared environment.
– GxE inflates heritability estimates when
environmental risk factors are shared by dizygotic
twins. Looks like the impact of their genes alone,
but is actually impact of GxE.
• Shared environment would include:
– Prenatal risk factors
– Widespread risk factors
G x E: One genotype is more sensitive to environment

Haynes 2003 EHP

 epiG E
II. Epigenetic modifications
(de novo or inherited)
• Epigenetic tags add to apparent heritability.
• Epigenetic tags may be set by prenatal or
other environment.
• May be de novo or possibly inherited
• Epigenetic tags may be amenable to change
via environmental interventions (e.g. diet).
(e.g., see Corrales 2009 Autism File magazine)
 G E
III. Genome damaged by environment
• Environmental factors can cause DNA damage (point
mutations or structural variations such as copy
number variants, CNVs, due to duplications or
deletions)
• This initially appears to be a de novo genetic cause of
autism.
• If inherited by subsequent generations it appears to
be an inherited genetic cause of autism.
• The root cause actually would be environmental and
potentially preventable, not genetic, for such cases.
• Unclear if mutation or CNV rates have risen.
• Signatures of mechanism may be informative (type of
SNP, or NHEJ vs. NAHR, FoSTeS, etc.). ROS
causes DSBs which precede NHEJ and possibly
NAHR. (Wenli Gu et al., 2008 PathoGenetics)
 G E
IV. G-E correlation:
Genes can affect exposure to
environmental factors
• Genes that affect non-shared environmental
exposures increase apparent heritability, but
prevention through environmental interventions may
be possible in such cases.
• Non-shared exposures driven by genes are those
where one DZ twin is more exposed than other, due
to their genotype.
• G-> Pica (tendency to put soil, etc. in mouth, causing
exposure to lead, etc.)
• G-> Dietary preferences (nutritional deficiency?)
• G-> Preference for social interaction, eye contact,
etc.
 Historical trends in exposure or
environmental interventions are hidden in
heritability estimates if GxE exists

• Irony: If GxE is important, growing environmental

exposures will increase disease prevalence, but
make heritability estimates higher over time.
Apparent impact of environment is reported as
smaller and smaller, even though in reality it is
growing.
• Twin studies only examine differences attributable to
environmental conditions that differ between twins,
but some environmental exposures are ubiquitous. If
some interventions have not been tried in the studied
twins, the potential impacts of those may not be
seen.
 Recap: How can environment be
important if heritability is so high?
I. GxE (interaction)

III. epiG E (epigenetic changes)

V. G E (genetic damage)

VII. G E (correlation)
 Why so little research on
environmental factors in autism?
• All autism research limited
(until recently)
• Behavioral > molecular level
(until recently)
• Genes > Environment:
High heritability
misinterpreted as
“genes not environment”
• Toxicology and epidemiology not sufficiently
integrated with autism research so far
 Toxicology literature and autism
literature have almost no overlap
Published 1985-2007, in PubMed:

– Toxicology literature: 1.5 million articles

– Autism literature: 10 thousand
– Overlap: Just 690 articles…
– Mostly on drug toxicity,
not about environmental chemicals
– Perhaps 2-4% of ASD literature
is on chemicals as risk factors?
See also: Herbert, Ringer, Corrales IMFAR 2008 poster #2979
 Integrating autism with
toxicology or epidemiology literature
One can seek potential risk factors using this logic:

ASD - Symptom - Risk Factor

Prepulse
AUTISM PRENATAL
inhibition INFECTION
deficit

often
can be
includes
caused by
ASD - Comorbidity - Risk Factor

AUTISM PRENATAL
ADHD ETS

is often
has as
comorbid
risk factor...
with...

In this type of analysis, one identifies known risk factors for diseases
that are often comorbid with or autism. The reasoning here is that
prenatal environmental tobacco smoke, alcohol, lead, or PCBs might be
risk factors for autism because they have been associated with ADHD.
 ASD studies rarely investigate
environmental risk factors
MOST VERY VERY
STUDIES FEW FEW

AUTISM AUTISM AUTISM AUTISM AUTISM

Behavior or
Neuropsych
trait
Brain
region

Cell

Gene Toxin
 Much literature linking
environmental toxins to autism-related
features, just not to autism per se

AUTISM
AUTISM

Behavior or
Neuropsych
Brain
trait
region
Cell
Biomarkers
Gene

Toxins
ASD – Gene – Toxin

AUTISM

Gene
Expression Toxins
 Search for Chemicals that Interact with
Autism Gene Candidates
• Objective, comprehensive search for risk
factors?
• 142 genes (autism candidates) selected
• Toxicogenomics database queried for reports of
chemicals “affecting” any of these genes
(e.g., gene expression changes in animal
studies)
• Approx. 500 chemicals were identified,
for 122 genes:
– xenobiotics (pollutants, pesticides, illegal drugs, etc.)
– medications
– nutrients
– endogenous substances
Corrales, IMFAR 2009 poster
 Results: Chemicals that Interact with
Autism Gene Candidates
Genes
reportedly interacting with
the most chemicals
• PON
( >120 chemicals)
• MET, PTEN, ADRB2, TH
(>30 each)
• MECP2, TSC2, RELN, UBE3A, GABRB3
(4-14 chemicals each)
• Most other genes
(1-2 chemicals so far)
 Results: Chemicals that Interact with
Autism Gene Candidates
Xenobiotics or related
chemicals
reportedly interacting with
the most candidate genes
Endogenous chemicals
reportedly interacting with
the most candidate genes
Dietary
• Carbon tetrachloride (33)
• t-butyl peroxide (19), hydrogen peroxide (9)
• Sodium arsenite (17), arsenic trioxide (8),
arsenic (4), nickel sulfate (9)
• LPS (11)
• Paraquat (10), chlorpyrifos (4)
• Benzene (7), B[a]P (7)
• Ethanol (7), tobacco smoke (4)
• BPA, TCDD
• Progesterone & estradiol
• Corticosterone
• Thyroxine
• Zinc
• Flavonoids
• Fats
 Recap and Next Steps
• Education about heritability of autism:
– GxE (interaction)
– epiG E (epi-mutation)
– G E (genetic damage)
– G E (correlation)
• Research on environmental factors
• Integrated study of E and G factors in autism
(e.g. what affects MECP2 expression?)
• Integration and prioritization across E findings in
autism literature (zinc, hormones, ROS, cytokines,
calcium, cholesterol, neurotransmitters, melatonin,
vitamin D, etc.)
• More integration with toxicology and epidemiology
from outside the autism literature, pulling in the
“autism-relevant” literature by
linking autism’s features to environment.
Thank you
 APPENDIX/
ADDITIONAL SLIDES
Population Attributable Fraction
(PAF)
What fraction of today’s cases would be avoided
if this risk factor were not present?
2.0

1.5

Extra risk
Risk

1.0
Baseline

0.5

0.0
e (unexposed) E (exposed)
News articles
Books with autism in title
 Autism funding, trials, research
• NIH research funding per DALY (disease adjusted life year)
(FY03-08):
– $150 for mental health
– $1,100 for cancer
– $2,800 for infectious diseases in the U.S.
• Federal grants for autism research (as of mid-2007):
– behaviors (almost 40% of grants)
– genetics (30-40% of grants)
• Autism research published 1985-2007:
– 20% on diagnosis
– 10% on therapy
– 1% on prevention
• 63 autism-related active intervention trials as of early 2008:
– Almost 75% (47 clinical trials) on medications
– 14% (9) on behavioral therapy
– 10% (6) on dietary therapy or supplements