2019 - Environmental Risk Factors and Biomarkers For Autism Spectrum Disorder - An Umbrella Review of The Evidence
2019 - Environmental Risk Factors and Biomarkers For Autism Spectrum Disorder - An Umbrella Review of The Evidence
2019 - Environmental Risk Factors and Biomarkers For Autism Spectrum Disorder - An Umbrella Review of The Evidence
Summary
Lancet Psychiatry 2019; Background Numerous studies have identified potential risk factors and biomarkers for autism spectrum disorder. We
6: 590–600 aimed to study the strength and validity of the suggested environmental risk factors or biomarkers of autism spectrum
See Comment page 551 disorder.
*Contributed equally
Yonsei University College of Methods We did an umbrella review and systematically appraised the relevant meta-analyses of observational studies.
Medicine, Seoul, Republic of We searched PubMed, Embase, and the Cochrane Database of Systematic Reviews for papers published between
Korea (J Y Kim, M J Son MD);
Department of Psychological
database inception and Oct 17, 2018, and screened the reference list of relevant articles. We obtained the summary
& Brain Sciences, Washington effect, 95% CI, heterogeneity, and 95% prediction intervals. We examined small study effects and excess significance.
University in St. Louis, MO, We did analyses under credibility ceilings. This review is registered with PROSPERO, number CRD42018091704.
USA (C Y Son BA); Early
Psychosis: Interventions and
Clinical-detection (EPIC) Lab, Findings 46 eligible articles yielded data on 67 environmental risk factors (544 212 cases, 81 708 787 individuals) and
Department of Psychosis 52 biomarkers (15 614 cases, 15 433 controls). Evidence of association was convincing for maternal age of 35 years or
Studies, Institute of Psychiatry, over (relative risk [RR] 1·31, 95% CI 1·18–1·45), maternal chronic hypertension (odds ratio [OR] 1·48, 1·29–1·70),
Psychology & Neuroscience,
maternal gestational hypertension (OR 1·37, 1·21–1·54), maternal overweight before or during pregnancy (RR 1·28,
King’s College London, London,
UK (J Radua MD); FIDMAG 1·19–1·36), pre-eclampsia (RR 1·32, 1·20–1·45), prepregnancy maternal antidepressant use (RR 1·48, 1·29–1·71),
Germanes Hospitalaries, and maternal selective serotonin reuptake inhibitor (SSRI) use during pregnancy (OR 1·84, 1·60–2·11). Only
CIBERSAM, Barcelona, Spain two associations, maternal overweight before or during pregnancy and SSRI use during pregnancy, retained their
(J Radua); Centre for Psychiatry
high level of evidence under subset sensitivity analyses. Evidence from biomarkers was scarce, being supported by
Research, Department of
Clinical Neuroscience, p values close to the significance threshold and too few cases.
Karolinska Institute,
Stockholm, Sweden (J Radua); Interpretation Convincing evidence suggests that maternal factors, such as age and features of metabolic syndrome,
Institut d’Investigacions
are associated with risk of autism spectrum disorder. Although SSRI use during pregnancy was also associated with
Biomèdiques August Pi i
Sunyer (IDIBAPS), Barcelona, such risk when exposed and non-exposed groups were compared, this association could be affected by other
Spain (J Radua); Department of confounding factors, considering that prepregnancy maternal antidepressant use was also convincingly associated
Pediatrics, Luton & Dunstable with higher risk of autism spectrum disorder. Findings from previous studies suggest that one possible confounding
University Hospital NHS
Foundation Trust, Luton, UK
factor is underlying maternal psychiatric disorders.
(M Eisenhut MD); CESP, Inserm
UMR1178, Department of Funding None.
Psychiatry, Assistance
Publique-Hôpitaux de Paris,
Bicêtre University Hospital,
Copyright © 2019 Elsevier Ltd. All rights reserved.
Le Kremlin Bicêtre, France
(F Gressier MD); Research and Introduction Advances have been made in the knowledge of genetic
Development Unit, Parc Autism spectrum disorder is a leading cause of disability causes of autism spectrum disorder; however, the exact
Sanitari Sant Joan de Déu,
Universitat de Barcelona,
in children, and often requires high levels of support, genes have not yet been elucidated. In addition, the
Fundació Sant Joan de Déu, which is costly for society and places a substantial results on associations of various kinds of environmental
Barcelona, Spain economic, emotional, and physical burden on affected factors for autism spectrum disorder have been
(A Koyanagi MD); Instituto de families.1–4 The prevalence of autism spectrum disorder inconsistent, hierarchies of evidence have not been
Salud Carlos III, Centro de
Investigación Biomédica en
was estimated to be 2·47% in US children and adolescents determined across different factors, and it is unknown
Red de Salud Mental, in 2014–165 and 7·6 per 1000 individuals globally in 2010, whether these risk factors are prone to bias.
CIBERSAM, Madrid, Spain when it accounted for 111 disability-adjusted life-years per Cohort and case-control studies have reported various
(A Koyanagi); Centre for 100 000 global population.2 types of environmental risk factors or biomarkers of
Addiction & Mental Health,
Toronto, ON, Canada
Given the scarcity of clinical and epidemiological autism spectrum disorder, and these have been meta-
(A F Carvalho MD); Department evidence of remission in autism spectrum disorder,2 analysed by combining the results of multiple studies.12
of Psychiatry, University of numerous investigations are focused on better under However, these analyses are usually restricted to one
Toronto, Toronto, ON, Canada
standing and advancing risk prediction and prevention of topic, and assessment of various kinds of bias in the
(A F Carvalho); Physiotherapy
Department, South London the disorder. The cause of autism spectrum disorder is literature is often insufficient. Claimed significant
and Maudsley NHS Foundation multifactorial, with various genetic predispositions and associations are susceptible to bias such as publication
Trust, London, UK environmental risk factors having been associated with bias, reporting bias, and residual confounding bias,
(B Stubbs PhD); Department of
an increased risk of autism spectrum disorder.6–11 resulting in false positives13 or inflated estimates14 of the
Psychological Medicine,
Research in context Institute of Psychiatry,
Psychology and Neuroscience,
Evidence before this study Added value of this study King’s College London, London,
We searched PubMed, Embase, and Cochrane Database of We identified and analysed 119 unique associations of UK (B Stubbs); Department of
Systematic Reviews between database inception and environmental risk factors or biomarkers with risk of autism Neurosciences and
Neurosciences Center,
Oct 17, 2018, for meta-analyses of observational studies spectrum disorder. Among these, only maternal factors, namely University of Padua, Padua,
studying any environmental risk factors and biomarkers of advanced age, chronic hypertension, pre-eclampsia, gestational Italy (M Solmi MD); Early
autism spectrum disorder, without language limitations, using hypertension, and overweight before or during pregnancy, were Psychosis: Interventions and
the following search terms: “autis*”, “Asperge*”, and convincingly associated with an increased risk of autism Clinical-detection (EPIC) Lab,
Department of Psychosis
“meta-analysis.” Our search showed that numerous risk factors spectrum disorder. Selective serotonin reuptake inhibitor use Studies, Institute of Psychiatry,
and biomarkers were associated with risk of autism spectrum during pregnancy was also convincingly associated with an Psychology & Neuroscience,
disorder in systematic reviews and meta-analyses. However, increased risk of autism spectrum disorder, but confounding King’s College London, London,
some results have been inconsistent, and it is unclear if the from underlying maternal psychiatric disorder is possible. UK (M Solmi, P Fusar-Poli MD);
Department of Psychiatry,
claimed associations are prone to biases in the literature. Evidence from biomarkers was scarce, supported by few cases University of Toledo Medical
One systematic review by Modabbernia and colleagues has and p values close to the significance threshold. Center, Toledo, Ohio, USA
comprehensively identified and analysed possible (T B Rais MD); Department of
Implications of all the available evidence Pediatrics, Yonsei University
environmental risk factors of autism spectrum disorder and
Our findings suggest that offspring of mothers who are older, College of Medicine, Seoul,
concluded that birth complications accompanied by trauma or South Korea (K H Lee MD,
have features of the metabolic syndrome, and might have
ischemia and hypoxia have strong associations with the J I Shin MD); Department of
psychiatric disorders are at an increased risk of developing
prevalence of autism spectrum disorder, but overall, Pediatrics, Severance Children’s
autism spectrum disorder. Although this finding does not imply Hospital, Seoul, South Korea
quantitative analysis was scarce and bias assessment was
that the other environmental risk factors and biomarkers we (KH Lee, JI Shin); Department of
incomplete owing to its reliance on previous reports. Internal Medicine IV, Medical
examined are not meaningful, there is still some uncertainty in
To overcome these limitations, we did an umbrella review of University Innsbruck,
their associations that should be resolved. Well-designed
meta-analyses. We did various tests of bias assessment and Anichstraße 35,
prospective cohort studies are needed to draw firmer 6020, Innsbruck, Austria
applied criteria for determining the level of credibility of the
conclusions. (A Kronbichler PhD); Pain and
association. Rehabilitation center and
Department of Medicine and
Health Sciences (IMH), Faculty
association. Such problems have resulted in an excess We included meta-analyses of observational studies of Health Sciences University of
Linköping, Linköping, Sweden
of significant associations (p<0·05) in psychological examining associations between autism spectrum dis (E Dragioti PhD); and OASIS
science and other medical fields.15,16 One systematic order and potential environmental risk factors or bio Service, South London and
overview7 has comprehensively identified and analysed markers. The definition of autism spectrum disorder Maudsley NHS Foundation
possible environmental risk factors of autism spectrum followed that of the original meta-analysis, whereas Trust, London, UK (P Fusar-Poli)
disorder. Although this review was informative, the risk factors and biomarkers were defined according to Correspondence to:
Dr Paolo Fusar-Poli, Department
quantitative assessment of bias was incomplete because WHO.17,18 Biomarkers were defined as any substance,
of Psychosis Studies, Institute of
it relied on reports from the original studies, and structure, or process that can be measured in the body or Psychiatry, Psychology &
definite criteria for determining the credibility of the its products that can influence or predict the incidence Neuroscience, London SE5 8AF,
associations were absent. To overcome these limitations, of outcome or disease.17 Risk factors were defined as any UK
paolo.fusar-poli@kcl.ac.uk
we did an umbrella review of the relevant meta-analyses. attribute, characteristic, or exposure of an individual
or
We aimed to generate a hierarchy of evidence and that increases the likelihood of developing a disease or
examine the true association of the suggested environ injury.18 Dr Jae Il Shin, Department of
Pediatrics, Yonsei University
mental risk factors and biomarkers with autism We screened for articles without language restriction. College of Medicine, Seoul
spectrum disorder. We only included meta-analyses that reported either 03722, South Korea
effect estimates of individual study estimates or the data shinji@yuhs.ac
Methods necessary to calculate these. When two or more meta-
Literature search strategy and eligibility criteria analyses existed for an association, we included the most
Three investigators (JYK, MJS, and CYS) searched recent meta-analysis with the largest number of studies.
PubMed, Embase, and the Cochrane Database of This review is registered with PROSPERO, number
Systematic Reviews for papers published between data CRD42018091704, and the protocol is available online.19
base inception and Oct 17, 2018, using the search terms
(Asperge* [All Fields] OR autis* [All Fields]) AND meta Data extraction
[All Fields]. We chose eligible articles by consecutively From each meta-analysis, we extracted the first author,
examining the titles, abstracts, and then the full-text publication year, risk factor or biomarker of interest,
(figure 1). We manually searched the references of the number of autism spectrum disorder cases and all
relevant articles and attempted to identify and include participants, maximally adjusted individual study estim
eligible studies. Disagreements were resolved via dis ates and corresponding 95% CI, metrics used for
cussion between JYK, MJS, CYS, and JIS. analyses—such as mean difference, Hedges’ g, Cohen’s d,
Factors listed are associated with increased risk of autism spectrum disorder, except the not significant row. For factors shown as a comparison (ie, x vs y), the former (ie, x) is associated with increased risk of
autism spectrum disorder compared with the latter (ie, y). Only two factors, maternal breastfeeding and maternal folic acid supplement during pregnancy, were associated with decreased risk of autism spectrum
disorder (not shown in this table). Both were graded as weak evidence (class IV). Heterogeneity was assessed in terms of Cochran’s Q test and large heterogeneity was defined as an I2 statistic of >50%.
Small study effects were assessed by Egger’s asymmetry test and were claimed at an Egger p value of <0·1 with estimate of the largest component study more conservative than summary estimate under random
effects model. Excess significance bias was claimed at p<0·1, with the observed number of significant studies larger than the expected number of significant studies. All statistical tests are two-sided.
PM10=particulate matter with a diameter of <10 µm. PM2·5=particulate matter with a diameter of <2·5 µm. SSRI=selective serotonin reuptake inhibitor. *The paternal age ≤35 years group has advanced age
compared with the reference group. †The meta-analysis supporting the association included studies by Geier and colleagues, for which concerns have been raised about the objective nature of the study
population. When study estimates from Geier and colleagues were excluded, the random effects summary estimate of the association changed to 0·98 (0·89–1·07), suggesting no association between thimerosal
exposure during embryo and early infancy with autism spectrum disorder.
Table 1: Summary of level of evidence for associations between environmental factors and risk of autism spectrum disorder
disorder according to important factors, such as sex both cohort and case-control design studies, eight (12%)
differences and presence of intellectual disability. used cohort design, six (9%) used case-control designs,
and six (9%) included cross-sectional studies. The median
Role of the funding source number of study estimates in each analysis was eight
There was no funding source for this study. All authors (range 2–24). The median number of cases was 3764 and
had full access to all of the study data and the corres the median total population was 502 843.
ponding authors had the final responsibility for the 52 (78%) of 67 associations were significant under the
decision to submit for publication. random effects model, of which 33 (49%) were p<1 × 10–³
and 18 (27%) were p<1 × 10–⁶. 52 (78%) associations had
Results more than 1000 autism spectrum disorder cases, of
1699 potentially eligible articles were identified by our which 16 were p<1 × 10–⁶. Of 52 significant associations,
initial search (figure 1). During the screening process, 40 were also supported by the significant result of the
14 articles were excluded because larger meta-analyses largest component study. Metrics were consistent with
were available (appendix pp 3–4). The eligible 46 articles12,41–85 those of the original meta-analyses except for one
yielded 119 associations (67 environmental risk factors and association (extremely low birthweight vs normal
52 biomarkers). 67 associations of environmental risk birthweight), where we converted Cohen’s d to OR;
factors with autism spectrum disorder were based on data ultimately, we used either RR or OR for all associations.
of 544 212 cases and a population of 81 708 787 (tables 1, 2; Effect size was smaller than 2 for all but seven
appendix pp 5–13). 42 (63%) of 67 associations included associations, of which three (congenital cytomegalovirus
Table 2: Association of environmental factors with risk of autism spectrum disorder graded convincing evidence (class I) or highly suggestive evidence (class II)
Standard error
autism spectrum disorder. 0·3
31 (46%) of 67 associations showed large heterogeneity 0·4
(I²>50%). 24 (36%) significant associations had neither
0·5
small study effects nor excess significance bias.
95% prediction intervals excluded the null value in only 0·6
evidence (tables 1, 2). The risk factors with convincing 4 Reference line (y=x)
associations were maternal age of 35 years or older versus
3
25–29 years (RR 1·31, 95% CI 1·18–1·45), maternal chronic
hypertension (OR 1·48, 1·29–1·70), maternal gestational 2
hypertension (OR 1·37, 1·21–1·54), maternal overweight
1
before or during pregnancy (RR 1·28, 1·19–1·36),
pre-eclampsia (RR 1·32, 1·20–1·45), prepregnancy mat 0
ernal anti depressant use (RR 1·48, 1·29–1·71), and
–1
selective serotonin reuptake inhibitor (SSRI) use during –1·0 –0·5 0 0·5 1·0 1·5 2·0 2·5 3·0 3·5 4·0
pregnancy (OR 1·84, 1·60–2·11). Eight (12%) associations Log (summary estimate under random effects model)
were graded as highly suggestive evidence (tables 1, 2).
Figure 3: Effect size of the largest study vs summary effect under random
11 (16%) associations were graded as suggestive evidence
effects model for each meta-analysis of environmental risk factors
(appendix p 5), 26 (39%) were graded as weak evidence The three outliers with a log of the summary estimate that is greater than 2 are
(appendix pp 6–7), and the remaining 15 (22%) did not associations of autism spectrum disorder with congenital cytomegalovirus
show statistically significant associations (appendix infection, hearing impairment, and visual impairment.
pp 8–9).
52 associations of biomarkers comprised 15 614 cases ceilings. 95% prediction intervals excluded the null value
and 15 433 controls (panel, appendix pp 14–20). Of 52 meta- in only one association (ratio of index finger length to ring
analyses of biomarkers, 17 (33%) used case-control studies. finger length [D2:D4 ratio]). Detailed results are shown in
Two (4%) associations used cross-sectional studies, and the appendix (pp 14–20).
study design was not specified in 33 (63%) studies. The We did a sensitivity subset analyses on meta-analyses of
median number of study estimates in each analysis was six 15 environmental risk factors graded as convincing
(range 2–23). The median number of cases was 228 and (class I) or highly suggestive (class II) evidence. Subset
the median number of controls was 216. 27 (52%) analysis restricted to cohort studies (prospective or
associations were significant, and ten (19%) associations retrospective) showed that only two associations of class I
were p<1 × 10–³. Moreover, only three associations—brain- remained at the same rank (appendix p 21). These
derived neurotrophic factor in blood, mercury in hair, and were maternal overweight before or during pregnancy
mercury in whole blood—were supported by a population and maternal pre-eclampsia. Three associations (SSRI use
with more than 1000 autism spectrum disorder cases. No during pregnancy, acetaminophen use during pregnancy,
associations were based on more than 1000 cases and and paternal age >45 years vs reference group) remained
p<1 × 10–³. Thus, for its level of evidence, no biomarker as highly suggestive evidence. When subset analysis was
association was graded as suggestive (class III) or above. restricted to only prospective cohort studies, no convincing
Of 27 statistically significant associations, only 14 were also association was seen, and only two associations (maternal
supported by a significant result of the largest component overweight before or during pregnancy and SSRI use
study. 44 associations (85%) had large heterogeneity during pregnancy) were still graded as highly suggestive
(I²>50%), of which 36 (69%) associations had very large evidence (appendix p 22).
heterogeneity (I²>75%). Only 11 (21%) associations In subset analyses of adjusted study estimates,
retained statistical significance under 10% credibility the association of maternal pre-eclampsia with autism
compared with mothers whose children showed typical (OR 1·81, 95% CI 1·44–2·29), supporting the idea that
development,92 in children with severe autism spectrum presence of a maternal psychiatric condition is an
disorder, autism spectrum disorder-specific autoantibodies independent risk factor for autism spectrum disorder.53
were significantly more prevalent in mothers with diabetes When SSRI-exposed groups were compared with SSRI-
(type 2 or gestational) and mothers who were moderately unexposed groups with a history of affective disorder (a
overweight compared with mothers without these setting in which the possibility of confounding by
conditions.91 Jones and colleagues90 showed that autism psychiatric disorder is minimised), the association with
spectrum disorder-specific antigen-induced maternal autism spectrum disorder was not significant (RR 1·18,
autoantibodies altered various autism spectrum disorder- 95% CI 0·91–1·52).64 Analyses restricted to sibling studies
relevant behaviours in mice. Therefore, one hypothesis is also showed a non-significant association (RR 0·96,
that metabolic syndrome could contribute to the production 95% CI 0·65–1·42).64 Notably, when individuals with
of autism spectrum disorder-specific maternal autoanti paternal antidepressant exposure during the maternal
bodies through the breakdown of maternal immune pregnancy period were compared with an unexposed
tolerance, which can increase the risk of developing autism group, the former group had an increased risk of autism
spectrum disorder in the offspring. spectrum disorder (RR 1·29, 95% CI 1·08–1·53).64 Overall,
Convincing evidence showed that maternal age, when these findings suggest that maternal psychiatric disorder
the comparison is restricted to age groups of 35 years or might act as an inde pendent risk factor for autism
older versus 25–29 years, was associated with an increased spectrum disorder and the association between SSRI use
risk of autism spectrum disorder. Accumulation of during pregnancy and autism spectrum disorder needs to
mutations, high rate of complications, and increased be further verified in adequately designed future studies.
chance of exposure to medications or pollutions could Maternal autoimmune disease was associated with
underlie the increased risk of autism spectrum disorder an increased risk of autism spectrum disorder, graded
in the advanced maternal age group.80 Advanced paternal as having a highly suggestive association, with the
age was also associated with an increased incidence of 95% prediction interval excluding the null value. In
autism spectrum disorder. Three comparisons—per mothers with autoimmune diseases, immune response
10-year increase in paternal age, highest paternal age mediators and autoantibodies might play a role in fetal
group vs reference group, paternal age >45 years vs neurodevelopment, resulting in adverse fetal outcomes
reference group, and paternal age 40–45 years vs reference such as autism spectrum disorder. Family history of
group—showed sufficiently low p values (<1 × 10–⁶) and psoriasis, rheumatoid arthritis, type 1 diabetes, or any
95% prediction intervals excluded the null value despite type of autoimmune disease was also associated with an
high heterogeneity and presence of small study effects. increased risk of autism spectrum disorder, graded as
In two of the comparisons (paternal age >45 years and suggestive evidence. Potential links between the
paternal age 40–45 years vs reference group), subset production of autism spectrum disorder-specific brain-
analyses of prospective studies showed p values of less reactive maternal autoantibodies and maternal auto
than 1 × 10–³ with no evidence of small study effects, immunity might underlie this association (appendix
indicating the existence of meaningful associations. An p 29).94,95 However, in the identified meta-analyses, small
increased rate of de novo mutations93 and epigenetic study effects existed across the associations; therefore,
alternation80 could underlie the association. more well conducted studies are needed to confirm the
Convincing evidence showed that maternal SSRI use association between maternal autoimmune disease and
during pregnancy was associated with a higher risk of autism spectrum disorder.
autism spectrum disorder than that of unexposed groups. We identified and analysed 52 biomarkers for autism
However, this association must be interpreted carefully. In spectrum disorder. Identifying robust evidence of
another meta-analysis, when maternal groups with biomarkers for autism spectrum disorder can result in
prepregnancy antidepressant exposure were compared early diagnosis and better treatment of the disease.96 The
with unexposed maternal groups, the associ ation with association of the D2:D4 ratio with a risk of autism
autism spectrum disorder was also graded as convincing spectrum disorder was supported by p<1 × 10–⁶ without
evidence. This finding raises the question of whether any signs of bias, meeting the criteria for convincing
underlying psychiatric conditions of mothers have caused evidence—except for the number of autism spectrum
confounding by indication in classical comparisons (SSRI disorder cases, being supported by 277 cases. However,
exposed vs SSRI unexposed). Several other meta- most of the associations of biomarkers were supported
analyses53,64 have attempted to discern between the two by p values close to the significance threshold (p>1 × 10–³)
possible causes of autism spectrum disorder. When and too few cases, implying the possibility of false
maternal groups with a psychiatric disorder but no SSRI positive findings. Similar findings were seen in umbrella
use during pregnancy were compared with maternal reviews of biomarkers for other psychiatric disorders.21,97,98
groups without a psychiatric disorder and also not exposed Although our review provided more strict and objective
to SSRIs during pregnancy, an increased incidence of criteria for assessing the associations compared with
autism spectrum disorder was seen in the former group previous reviews (appendix p 29), our study has some
18 WHO. Health topics: risk factors. http://www.who.int/topics/risk_ 44 Chen SW, Zhong XS, Jiang LN, et al. Maternal autoimmune diseases
factors/en/ (accessed Sept 30, 2018). and the risk of autism spectrum disorders in offspring: a systematic
19 PROSPERO. An umbrella review of meta-analyses on risk factors for review and meta-analysis. Behav Brain Res 2016; 296: 61–9.
autism spectrum disorder. http://www.crd.york.ac.uk/PROSPERO/ 45 Dachew BA, Mamun A, Maravilla JC, Alati R. Pre-eclampsia and the
display_record.php?ID=CRD42018091704 (accessed Sept 30, 2018). risk of autism-spectrum disorder in offspring: meta-analysis.
20 Belbasis L, Köhler CA, Stefanis N, et al. Risk factors and peripheral Br J Psychiatry 2018; 212: 142–47.
biomarkers for schizophrenia spectrum disorders: an umbrella 46 Do B, Lynch P, Macris EM, et al. Systematic review and meta-analysis
review of meta-analyses. Acta Psychiatr Scand 2018; 137: 88–97. of the association of autism spectrum disorder in visually or hearing
21 Bellou V, Belbasis L, Tzoulaki I, Middleton LT, Ioannidis JPA, impaired children. Ophthalmic Physiol Opt 2017; 37: 212–24.
Evangelou E. Systematic evaluation of the associations between 47 Flores-Pajot MC, Ofner M, Do MT, Lavigne E, Villeneuve PJ.
environmental risk factors and dementia: an umbrella review of Childhood autism spectrum disorders and exposure to nitrogen
systematic reviews and meta-analyses. Alzheimers Dement 2017; dioxide, and particulate matter air pollution: a review and
13: 406–18. meta-analysis. Environ Res 2016; 151: 763–76.
22 Bellou V, Belbasis L, Tzoulaki I, Evangelou E, Ioannidis JPA. 48 Gabriele S, Sacco R, Persico AM. Blood serotonin levels in autism
Environmental risk factors and Parkinson’s disease: an umbrella spectrum disorder: a systematic review and meta-analysis.
review of meta-analyses. Parkinsonism Relat Disord 2016; 23: 1–9. Eur Neuropsychopharmacol 2014; 24: 919–29.
23 Belbasis L, Bellou V, Evangelou E, Ioannidis JP, Tzoulaki I. 49 Honekopp J. Digit ratio 2D:4D in relation to autism spectrum
Environmental risk factors and multiple sclerosis: an umbrella disorders, empathizing, and systemizing: a quantitative review.
review of systematic reviews and meta-analyses. Lancet Neurol 2015; Autism Res 2012; 5: 221–30.
14: 263–73. 50 Jafari T, Rostampour N, Fallah AA, Hesami A. The association
24 He Y, Li X, Gasevic D, et al. Statins and multiple noncardiovascular between mercury levels and autism spectrum disorders: a systematic
outcomes: umbrella review of meta-analyses of observational studies review and meta-analysis. J Trace Elem Med Biol 2017; 44: 289–97.
and randomized controlled trials. Ann Intern Med 2018; 169: 543–53. 51 Jiang HY, Xu LL, Shao L, et al. Maternal infection during pregnancy
25 Kyrgiou M, Kalliala I, Markozannes G, et al. Adiposity and cancer at and risk of autism spectrum disorders: a systematic review and
major anatomical sites: umbrella review of the literature. BMJ 2017; meta-analysis. Brain Behav Immun 2016; 58: 165–72.
356: j477. 52 Jung Y, Lee AM, McKee SA, Picciotto MR. Maternal smoking and
26 Sterne JAC, Davey Smith G. Sifting the evidence-what’s wrong with autism spectrum disorder: meta-analysis with population smoking
significance tests? BMJ 2001; 322: 226–31. metrics as moderators. Sci Rep 2017; 7: 4315.
27 Ioannidis JPA, Tarone R, McLaughlin JK. The false-positive to 53 Kaplan YC, Keskin-Arslan E, Acar S, Sozmen K. Maternal SSRI
false-negative ratio in epidemiologic studies. Epidemiology 2011; discontinuation, use, psychiatric disorder and the risk of autism in
22: 450–56. children: a meta-analysis of cohort studies. Br J Clin Pharmacol 2017;
28 Cochran WG. The combination of estimates from different 83: 2798–806.
experiments. Biometrics 1954; 10: 101–29. 54 Lam J, Sutton P, Kalkbrenner A, et al. A systematic review and
29 Ioannidis JPA, Patsopoulos NA, Evangelou E. Uncertainty in meta-analysis of multiple airborne pollutants and autism spectrum
heterogeneity estimates in meta-analyses. BMJ 2007; 335: 914–16. disorder. PLoS One 2016; 11: e0161851.
30 Higgins JPT, Thompson SG, Spiegelhalter DJ. A re-evaluation of 55 Li YM, Ou JJ, Liu L, Zhang D, Zhao JP, Tang SY. Association between
random-effects meta-analysis. J R Stat Soc Ser A Stat Soc 2009; maternal obesity and autism spectrum disorder in offspring:
172: 137–59. a meta-analysis. J Autism Dev Disord 2016; 46: 95–102.
31 Egger M, Davey Smith G, Schneider M, Minder C. Bias in 56 Liu L, Gao J, He X, Cai Y, Wang L, Fan X. Association between
meta-analysis detected by a simple, graphical test. BMJ 1997; assisted reproductive technology and the risk of autism spectrum
315: 629–34. disorders in the offspring: a meta-analysis. Sci Rep 2017; 7: 46207.
32 Ioannidis JPA, Trikalinos TA. An exploratory test for an excess of 57 Maeyama K, Tomioka K, Nagase H, et al. Congenital Cytomegalovirus
significant findings. Clin Trials 2007; 4: 245–53. infection in children with autism spectrum disorder: systematic
33 Ioannidis JPA. Clarifications on the application and interpretation review and meta-analysis. J Autism Dev Disord 2018; 48: 1483–91.
of the test for excess significance and its extensions. J Math Psychol 58 Main PA, Angley MT, O’Doherty CE, Thomas P, Fenech M.
2013; 57: 184–87. The potential role of the antioxidant and detoxification properties of
34 Salanti G, Ioannidis JPA. Synthesis of observational studies should glutathione in autism spectrum disorders: a systematic review and
consider credibility ceilings. J Clin Epidemiol 2009; 62: 115–22. meta-analysis. Nutr Metab 2012; 9: 35.
35 Papatheodorou SI, Tsilidis KK, Evangelou E, Ioannidis JPA. 59 Masarwa R, Levine H, Gorelik E, Reif S, Perlman A, Matok I.
Application of credibility ceilings probes the robustness of Prenatal exposure to acetaminophen and risk for attention deficit
meta-analyses of biomarkers and cancer risk. J Clin Epidemiol 2015; hyperactivity disorder and autistic spectrum disorder: a systematic
68: 163–74. review, meta-analysis, and meta-regression analysis of cohort
studies. Am J Epidemiol 2018; 187: 1817–27.
36 RStudio Team. RStudio: integrated development for R. Boston, MA:
RStudio, 2015. 60 Masi A, Quintana DS, Glozier N, Lloyd AR, Hickie IB, Guastella AJ.
Cytokine aberrations in autism spectrum disorder: a systematic
37 Viechtbauer W. Conducting meta-analyses in R with the metafor
review and meta-analysis. Mol Psychiatry 2015; 20: 440–46.
package. J Stat Softw 2010; 36: 1–48.
61 Mathewson KJ, Chow CH, Dobson KG, Pope EI, Schmidt LA,
38 Champely S, Ekstrom C, Dalgaard P, et al. pwr: basic functions for
Van Lieshout RJ. Mental health of extremely low birth weight
power analysis. R package version 1.2-2. March 3, 2018.
survivors: a systematic review and meta-analysis. Psychol Bull 2017;
https://cran.r-project.org/package=pwr (accessed Sept 30, 2018).
143: 347–83.
39 Ioannidis JPA. Integration of evidence from multiple
62 Mazahery H, Stonehouse W, Delshad M, et al. Relationship between
meta-analyses: a primer on umbrella reviews, treatment networks
long chain n-3 polyunsaturated fatty acids and autism spectrum
and multiple treatments meta-analyses. CMAJ 2009; 181: 488–93.
disorder: systematic review and meta-analysis of case-control and
40 Fusar-Poli P, Radua J. Ten simple rules for conducting umbrella randomised controlled trials. Nutrients 2017; 9: e155.
reviews. Evid Based Ment Health 2018; 21: 95–100.
63 Modabbernia A, Mollon J, Boffetta P, Reichenberg A. Impaired gas
41 Amin SB, Smith T, Wang H. Is neonatal jaundice associated with exchange at birth and risk of intellectual disability and autism:
autism spectrum disorders: a systematic review. J Autism Dev Disord a meta-analysis. J Autism Dev Disord 2016; 46: 1847–59.
2011; 41: 1455–63.
64 Morales DR, Slattery J, Evans S, Kurz X. Antidepressant use during
42 Andalib S, Emamhadi MR, Yousefzadeh-Chabok S, et al. pregnancy and risk of autism spectrum disorder and attention
Maternal SSRI exposure increases the risk of autistic offspring: deficit hyperactivity disorder: systematic review of observational
A meta-analysis and systematic review. Eur Psychiatry 2017; 45: 161–63. studies and methodological considerations. BMC Med 2018; 16: 6.
43 Babaknejad N, Sayehmiri F, Sayehmiri K, Mohamadkhani A, 65 Ng DK, Chan CH, Soo MT, Lee RS. Low-level chronic mercury
Bahrami S. The relationship between zinc levels and autism: exposure in children and adolescents: meta-analysis. Pediatr Int 2007;
a systematic review and meta-analysis. Iran J Child Neurol 2016; 49: 80–87.
10: 1–9.
66 Oldereid NB, Wennerholm UB, Pinborg A, et al. The effect of 83 Zheng Z, Zhang L, Zhu T, Huang J, Qu Y, Mu D.
paternal factors on perinatal and paediatric outcomes: a systematic Peripheral brain-derived neurotrophic factor in autism spectrum
review and meta-analysis. Hum Reprod Update 2018; 24: 320–89. disorder: a systematic review and meta-analysis. Sci Rep 2016;
67 Rutigliano G, Rocchetti M, Paloyelis Y, et al. Peripheral oxytocin and 6: 31241.
vasopressin: Biomarkers of psychiatric disorders? A comprehensive 84 Zheng Z, Zhu T, Qu Y, Mu D. Blood glutamate levels in autism
systematic review and preliminary meta-analysis. Psychiatry Res 2016; spectrum disorder: a systematic review and meta-analysis.
241: 207–20. PLoS One 2016; 11: e0158688.
68 Saghazadeh A, Rezaei N. Systematic review and meta-analysis links 85 Zhou XH, Li YJ, Ou JJ, Li YM. Association between maternal
autism and toxic metals and highlights the impact of country antidepressant use during pregnancy and autism spectrum
development status: Higher blood and erythrocyte levels for mercury disorder: an updated meta-analysis. Mol Autism 2018; 9: 21.
and lead, and higher hair antimony, cadmium, lead, and mercury. 86 Goines PE, Ashwood P. Cytokine dysregulation in autism spectrum
Prog Neuropsychopharmacol Biol Psychiatry 2017; 79: 340–68. disorders (ASD): possible role of the environment.
69 Sanchez CE, Barry C, Sabhlok A, et al. Maternal pre-pregnancy Neurotoxicol Teratol 36: 67–81.
obesity and child neurodevelopmental outcomes: a meta-analysis. 87 Faa G, Manchia M, Pintus R, Gerosa C, Marcialis MA, Fanos V.
Obes Rev 2018; 19: 464–84. Fetal programming of neuropsychiatric disorders.
70 Sandin S, Hultman CM, Kolevzon A, Gross R, MacCabe JH, Birth Defects Res C Embryo Today 2016; 108: 207–23.
Reichenberg A. Advancing maternal age is associated with 88 Korkmazer E, Solak N. Correlation between inflammatory markers
increasing risk for autism: a review and meta-analysis. and insulin resistance in pregnancy. J Obstet Gynaecol 2015;
J Am Acad Child Adolesc Psychiatry 2012; 51: 477–86. 35: 142–45.
71 Sayehmiri F, Babaknejad N, Bahrami S, Sayehmiri K, Darabi M, 89 Jones KL, Van de Water J. Maternal autoantibody related autism:
Rezaei-Tavirani M. Zn/Cu levels in the field of autism disorders: mechanisms and pathways. Mol Psychiatry 2019; 24: 252–65.
a systematic review and meta-analysis. Iran J Child Neurol 2015; 90 Jones KL, Pride MC, Edmiston E, et al. Autism-specific maternal
9: 1–9. autoantibodies produce behavioral abnormalities in an endogenous
72 Schur RR, Draisma LW, Wijnen JP, et al. Brain GABA levels across antigen-driven mouse model of autism. Mol Psychiatry 2018;
psychiatric disorders: A systematic literature review and published online June 28. DOI:10.1038/s41380-018-0126-1.
meta-analysis of (1) H-MRS studies. Hum Brain Mapp 2016; 91 Krakowiak P, Walker CK, Tancredi D, Hertz-Picciotto I,
37: 3337–52. Van de Water J. Autism-specific maternal anti-fetal brain
73 Taylor LE, Swerdfeger AL, Eslick GD. Vaccines are not associated autoantibodies are associated with metabolic conditions.
with autism: an evidence-based meta-analysis of case-control and Autism Res 2017; 10: 89–98.
cohort studies. Vaccine 2014; 32: 3623–29. 92 Krakowiak P, Walker CK, Bremer AA, et al. Maternal metabolic
74 Tseng PT, Chen YW, Stubbs B, et al. Maternal breastfeeding and conditions and risk for autism and other neurodevelopmental
autism spectrum disorder in children: a systematic review and disorders. Pediatrics 2012; 129: e1121–28.
meta-analysis. Nutr Neurosci 2019; 22: 354–62. 93 Sanders SJ, Murtha MT, Gupta AR, et al. De novo mutations
75 Wan H, Zhang C, Li H, Luan S, Liu C. Association of maternal revealed by whole-exome sequencing are strongly associated with
diabetes with autism spectrum disorders in offspring: a systemic autism. Nature 2012; 485: 237–41.
review and meta-analysis. Medicine 2018; 97: e9438. 94 Martin LA, Ashwood P, Braunschweig D, Cabanlit M,
76 Wang M, Li K, Zhao D, Li L. The association between maternal use of Van de Water J, Amaral DG. Stereotypies and hyperactivity in
folic acid supplements during pregnancy and risk of autism spectrum rhesus monkeys exposed to IgG from mothers of children with
disorders in children: a meta-analysis. Mol Autism 2017; 8: 51. autism. Brain Behav Immun 2008; 22: 806–16.
77 Wang T, Shan L, Du L, et al. Serum concentration of 95 Brimberg L, Sadiq A, Gregersen PK, Diamond B. Brain-reactive IgG
25-hydroxyvitamin D in autism spectrum disorder: a systematic correlates with autoimmunity in mothers of a child with an autism
review and meta-analysis. Eur Child Adolesc Psychiatry 2016; spectrum disorder. Mol Psychiatry 2013; 18: 1171–77.
25: 341–50. 96 Goldani AAS, Downs SR, Widjaja F, Lawton B, Hendren RL.
78 Wang Y, Tang S, Xu S, Weng S, Liu Z. Maternal body mass index Biomarkers in autism. Front Psychiatry 2014; 5: 100.
and risk of autism spectrum disorders in offspring: a meta-analysis. 97 Carvalho AF, Köhler CA, Brunoni AR, et al. Bias in peripheral
Sci Rep 2016; 6: 34248. depression biomarkers. Psychother Psychosom 2016; 85: 81–90.
79 Wu S, Ding Y, Wu F, et al. Family history of autoimmune diseases 98 Carvalho AF, Köhler CA, Fernandes BS, et al. Bias in emerging
is associated with an increased risk of autism in children: biomarkers for bipolar disorder. Psychol Med 2016; 46: 2287–97.
a systematic review and meta-analysis. Neurosci Biobehav Rev 2015;
99 Geier DA, Geier MR. A prospective study of thimerosal-containing
55: 322–32.
Rho(D)-immune globulin administration as a risk factor for autistic
80 Wu S, Wu F, Ding Y, Hou J, Bi J, Zhang Z. Advanced parental age disorders.J Matern Fetal Neonatal Med 2007; 20: 385–90.
and autism risk in children: a systematic review and meta-analysis.
100 Geier DA, Mumper E, Gladfelter B, Coleman L, Geier MR.
Acta Psychiatr Scand 2017; 135: 29–41.
Neurodevelopmental disorders, maternal Rh-negativity,
81 Xu RT, Chang QX, Wang QQ, et al. Association between hypertensive and Rho(D) immune globulins: a multi-center assessment.
disorders of pregnancy and risk of autism in offspring: a systematic Neuro Endocrinol Lett 2008; 29: 272–80.
review and meta-analysis of observational studies. Oncotarget 2018;
9: 1291–301.
82 Yoshimasu K, Kiyohara C, Takemura S, Nakai K. A meta-analysis of
the evidence on the impact of prenatal and early infancy exposures
to mercury on autism and attention deficit/hyperactivity disorder in
the childhood. Neurotoxicology 2014; 44: 121–31.