Systematic review of the association between particulate matter exposure and autism spectrum disorders

Environmental Research 153 (2017) 150–160

Contents lists available at ScienceDirect

Environmental Research journal homepage: www.elsevier.com/locate/envres

Review article

Systematic review of the association between particulate matter exposure and autism spectrum disorders María Morales-Suárez-Varelaa,b,c,

⁎,1

MARK

, Isabel Peraita-Costaa,1, Agustín Llopis- Gonzáleza,b,c,1

a Unit of Public Health, Hygiene and Environmental Health, Department of Preventive Medicine and Public Health, Food Science, Toxicology and Legal Medicine, University of Valencia, 46100 Valencia, Spain b CIBER Epidemiología y Salud Pública (CIBERESP), Instituto de Salud Carlos III, 28029 Madrid, Spain c Center for Advanced Research in Public Health (CSISP-FISABIO), 46010 Valencia, Spain

A R T I C L E I N F O

A BS T RAC T

Keywords: Particulate matter Autism ASD Environmental pollutants Systematic review

Particulate matter (PM) as an environmental pollutant is suspected to be associated with autism spectrum disorders. The aim of the present study was to review the epidemiological literature currently available on the relation between PM exposure and diagnosis of ASD. The PubMed database was searched from November 2015 up to January 2016 by one of the authors. We included observational studies (cohort and case–control studies) published in English carried out in children within the last 10 years, measuring PM exposure and health outcomes related to ASD. 13 studies met the inclusion criteria. Four of the studies found no association between PM exposure and ASD. The other 8 studies show positive associations restricted to specific exposure windows which however do not reach statistical significance at times. To conclude, the evidence from the studies allows us to conclude that there is an association between PM exposure and ASD whose strength varies according to the particle size studied with the association with PM2.5 and diesel PM being stronger. Given the potential importance for public health, cohort studies with proper adjustment for confounding variables and identification of critical windows of exposure are urgently needed to further improve knowledge about potential causal links between PM exposure and the development of ASD.

1. Introduction Autism spectrum disorders (ASD) is a group of heterogeneous neurodevelopmental disorders characterized by impairment in communication and social interaction accompanied with repetitive and restrictive behaviors (Lai et al., 2014). These symptoms present themselves in the early developmental period and can cause significant social and occupational impairment. ASD now encompasses the previous autistic disorder (autism), Asperger's disorder, childhood disintegrative disorder, and pervasive developmental disorder not otherwise specified (PDD-NOS). Globally speaking, the prevalence today of ASD is estimated to be between 6.2 and 7.6/1000 persons and rising (Elsabbagh et al., 2012; Baxter et al., 2015) which leads to a higher interest in the possible causes and risk factors associated with it. Up until recently, the study of the etiology of ASD was centered around its genetic component (Sandin et al., 2014; Lai et al., 2014; Geschwind et al., 2007; Newschaffer et al., 2007) and the potential environmental contributions were less investigated [Lawler, 2008].

⁎

1

Recent studies present evidence that suggest that environmental factors such as air pollution play a greater role as risk factors than previously thought (Hallmayer et al., 2011; Lyall et al., 2014; Grandjean et al., 2014; Sandin et al., 2014; Kalkbrenner et al., 2014). Environmental or air pollution can be defined as the mixture of gases and particles that contaminate the atmosphere and modify its natural characteristics (WHO, 2006). The emission of air pollution can be commonly traced back to the burning of fossil fuels as well as industrial and agricultural processes It has become a global public health concern and among the pollutants of most concern we find particulate matter (PM) (PM2.5, PM10 and diesel PM), carbon monoxide (CO), ozone (O3), nitrogen dioxide (NO2) and sulfur dioxide (SO2) (WHO 2005; WHO, 2013). A large number of studies support an association between air pollution and ASD; however specific evidence for individual constituents is conflicting or limited. The different components of air pollution could act synergistically and further studies on this phenomenon are needed. However, PM is a component of particular interest in relation to neurodevelopment and specifically ASD. PM components are suspected to be one of the major culprits of

Correspondence to: University of Valencia, Avda. Vicent Andrés Estellés, s/n, 46100 Burjassot, Valencia, Spain. E-mail address: [email protected] (M. Morales-Suárez-Varela). These authors contributed equally to this work.

http://dx.doi.org/10.1016/j.envres.2016.11.022 Received 21 June 2016; Received in revised form 28 November 2016; Accepted 29 November 2016 0013-9351/ © 2016 Elsevier Inc. All rights reserved.

Environmental Research 153 (2017) 150–160

M. Morales-Suárez-Varela et al.

the neurological effects of air pollution (Block et al., 2012) given that they may penetrate cellular membranes (Geiser et al., 2005; RothenRutishauser et al., 2008) and translocate from the systemic circulation or via the nasal mucosa and the olfactory bulb to the lungs and into the brain (Campbell et al., 2009; Oberdorster et al., 2009). Therefore a review of its stand-alone association with ASD is warranted. The data included here from the selected publications is that from the onepollutant models when data from both one and multi-pollutant models was available. PM is a very broad category, containing not only particles of different sizes, but also different sources and constituents. In this review PM will be divided into the following three categories; PM2.5, PM10 and diesel PM. Below this review will explore the possible relation between PM and ASD through a systematic review of the literature published on the possible relation among air pollution and ASD. 2. Methods 2.1. Study identification and eligibility criteria A preliminary search performed to assess the prevalence of other systematic reviews covering the possible association between PM2.5, PM10 and/or diesel PM and ASD yielded 4 relevant articles (Kalkbrenner et al., 2014; Rossignol et al., 2014; Suades-González et al., 2015 and Weisskopf et al., 2015) whose findings have been taken into account when discussing our own conclusions. The database we based our search on to identify publications eligible for inclusion in our review was PubMed, which was accessed between November 1, 2015 and January 1, 2016 for relevant studies, using the keywords: “environmental,” “pollution,” and “particulate” combined with “autism”. Fig. 1 shows the search strategy followed for this review. A total of 1063 articles were identified. Due to the large number of studies found the first item on the eligibility criteria (human study subjects) was used at this point to reduce the number of articles to be examined, which left n=717 articles. Initial screening identified 51 candidate studies. The initial screening of the studies was performed using the information available in the title and abstract. These potentially relevant studies were retrieved in full text and assessed for eligibility. The eligibility criteria used were that the study had to:

Fig. 1. Search strategy.

Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) recommendations (Friedenreich, 1993; Liberati et al., 2009).

1. Include humans as study subjects without restriction on the demographic characteristics of the population. 2. Conduct exposure assessment to PM2.5, PM10 and/or diesel PM during pregnancy or early childhood. 3. Include measures of autism symptoms or diagnosis. 4. Be a primary research article published after 2005.

2.2. Internal validity To further assess the chosen articles and guide the evaluation of the data included in them, we classified the publications using the scale proposed by the Scottish Intercollegiate Guidelines Network [SIGN, 2008] for establishing levels of evidence (Table 1) and recommendations (Table 2). Evidence is classified by its epistemologic strength and only the strongest gives way to strong recommendations while the weaker evidence can only yield weak recommendations. The scale proposes that the study design and the risk of bias be used to assess the quality of the scientific evidence provided (level of evidence). To rate the study design numbers (1−4) are used, while signs (++, + and -) are used to represent the assessed risk of bias. Based on this assessment of the quality of the evidence in the articles, grades (A–D) are then used to classify the strength of associated recommendations. This scale was chosen following the principles of evidence based medicine (EBM) which emphasizes the use of evidence from well designed and conducted research. EBM is about making sure that when a decision is made it is based on the most up to date, reliable and scientifically solid evidence available on the particular situation being studied (Sackett, 1997). Depending on the defined area of study, the quality or rating of the current best evidence available may be constrained due to ethical or other limitations. These restrictions on the rating of the evidence are at times insurmountable and must not be seen as detrimental to the study just as another characteristic of it. The

The publications were only included in the analysis if they met all the eligibility criteria. After a full assessment of the potentially relevant studies, initially 13 were proposed to be included in this systematic review. Of these, 12 included data relating specifically to PM exposure. 5 studies included data on PM2.5 (Guxens et al., 2015; Becerra et al., 2013; Talbott et al., 2015a; Raz et al., 2015 and Volk et al., 2013), 7 on PM10 (Jung et al., 2013; Gong et al., 2014; Guxens et al., 2015; Becerra et al., 2013, Kalkbrenner et al., 2015; Raz et al., 2015, and Volk et al., 2013) and 5 on diesel PM (Kalkbrenner et al., 2010; Talbott et al., 2015b; Windham et al., 2006; Roberts et al., 2013 and Volk et al., 2011). The remaining publication included indirect information based on the distance of the exposure area to freeways but after consideration was deemed apt to be included in this study. All of the 13 research articles included in this review have been included in at least one of the three previously mentioned reviews. However, the different approach with which this review has been conducted justifies the inclusion of these studies. This paper was written taking into account the methodological norms established for the publication of systematic reviews and the 151

Environmental Research 153 (2017) 150–160

M. Morales-Suárez-Varela et al.

4. Adjustment of results due to possible confounding factors.

Table 1 Levels of evidence.

The ideal study would include a sample size of tens of thousands, have clear inclusion and assessment criteria, use DSM-V as the ASD diagnostic criteria, measure PM locally and daily from conception through year two post birth and properly account for possible confounders. As is expected, none of the studies were classifiable as “perfect” however all are acceptable for inclusion and analysis. All of the studies included in this review were considered to have a large enough sample size, clear inclusion and assessment criteria and adequate ASD diagnosis criteria. However, different weights should be attributed to studies with vastly different sample sizes; a study of a sample size of tens of thousands of children is not equal to that of only a couple hundred. The difference in the ASD diagnostic criteria will be discussed later in the results. Exposure assessment and timing has also been considered adequate however, not all of the methods used are equal in their strength. It is not the same to asses PM exposure from NATA data which is based on estimates from multiple sources and then averaged annually than to use daily local measurements or to extrapolate exposure values according to residential proximity to roadways. All the studies take into account, in different degrees, the presence of confounders such as but not restricted to; maternal age, parental education level and season of birth. The adjusted results can be assumed to reflect the association between PM and ASD without the influence of the confounders each study has accounted for. We must mention however the possibility of having not properly accounted for confounders in those studies involving diesel PM since as a traffic related air pollutant it is part of a mixture of emissions where separating one component is virtually impossible. In these studies, it would be more appropriate to interpret diesel PM as a proxy of diesel emissions which would be the causal agent involved in the development of ASD if an association is found. In this review, the classification of the studies according to their level of evidence, grade of recommendation and internal validity serves first to evaluate the suitability of each study to be included for analysis and allows appropriately weighing of the studies in relation to one another. The classification according to the SIGN scale and assessment of internal validity was performed mainly by one author with frequent consultation with a second author and once a consensus among the two was reached, the classification was given to the third author for revision and approval. No notable disagreements arose among the authors at this point.

LE 1

1++ 1+

2

12++

2+

23 4

High-quality meta-analyses, systematic reviews of RCTs, or RCTs with a very low risk of bias Well-conducted meta-analyses, systematic reviews of RCTS, or RCTs with a low risk of bias Meta-analyses, systematic reviews, or RCTs with a high risk of bias High-quality systematic reviews of case-control or cohort or studies High-quality case-control or cohort studies with a very low risk of confounding or bias and a high probability that the relationship is casual Well-conducted case control or cohort studies with a low risk of confounding or bias and a moderate probability that the relationship is casual. Case control or cohort studies with a high risk of confounding or bias and a significant risk that the relationship is not causal Non-analytic studies, e.g. Case reports, case series. Expert opinion.

Abbreviations: SIGN, Scottish Intercollegiate Guidelines Network (2008); LE, levels of evidence; RCT: randomized and controlled trials. Table 2 Grades of recommendation. GR A

B

C

D

At least one meta-analysis, systematic review, or RCT rated as 1++, and directly applicable to the target population A body of evidence consisting principally of studies rated as 1+, directly applicable to the target population, and demonstrating overall consistency of results A body of evidence including studies rated as 2++, directly applicable to the target population, and demonstrating overall consistency of results Extrapolated evidence from studies rated as 1++, or 1+ A body of evidence including studies rated as 2+, directly applicable to the target population, and demonstrating overall consistency of results Extrapolated evidence from studies rated as 2++ Evidence level 3 or 4 Extrapolated evidence from studies rated as 2+

Abbreviations: SIGN, Scottish Intercollegiate Guidelines Network (2008). GR, Grade of Recommendation; RCT: randomized and controlled trials.

study of the association between ASD and PM does not lend itself to randomized clinical trials; ethical constraints limit the current best available evidence to case-control or cohort type studies. This proves to be a challenge when establishing an association between ASD and PM since it means that the included studies could at most receive a 2++ B rating and therefore the recommendations extracted from this review could at most be classified at moderately strong. However, given that the principles of EBM have been followed correctly the conclusions of these review are valid as they are derived from the best currently available evidence. All of the studies included in this review are case-control or cohort studies and therefore can only be scored as level 2 s. The assessed risk of bias and degree of probability that the relationship is casual, represented by the plus and minus signs places most studies in the 2+ category given their results (as presented in Table 3). The prevalent level 2+ rating of the studies included limits the strength of associated recommendations to grades C or D. When taking into account the target population and consistency of the results in each of the studies most end up categorized as having a grade of recommendation C. The internal validity of the studies was also taken into consideration when evaluating and summarizing the evidence. The factors affecting internal validity considered were:

2.3. Data extraction When compiling the results from the different studies, it was noticed that the results were expressed in a non-homogenous or standardized manner. In order to avoid any possible confusion, the results were standardized and presented in a single integrated scale. The available data and methodology utilized in the different studies restricted the standardization to a limited number of the studies (6/ 13). 3. Results 3.1. Characteristics of the studies The chosen studies were analyzed according the following characteristics: location, birth years, sample size, study design, ASD measurement criteria and classification, exposure assessment and timing, results, level of evidence and grade of recommendation. Table 3 summarizes the characteristics of the studies. The results presented in Tables 4–6 are adjusted results from the included studies. Tables 4, 5 and Figs. 2 and 3 show the ORs standardized to ΔPM=5 μg/ m3 for those articles (6 out of the 13 included) in which the ORs were presented for a given increase in PM concentration and for which

1. Presence of a large enough sample size. 2. Specification of inclusion and assessment criteria. 3. Quality of ASD diagnosis and exposure assessment.

152

153

USA

California, USA

California, USA

Windham et al., 2006

Roberts et al., 2013

Volk et al., 2013

Volk et al., 2011

1997–2006

1997–2006

1987–2002

1994

1990–2002

305/259

279/245

325/22,101

284/687

245/1767

979/14,666

Case-control

Case-control

Cohort

Nested casecontrol

Nested casecontrol

Nested casecontrol

Case-control

ADOS + ADI-R

ADOS + ADI-R

DSM-IV-R criteria applied to developmental evaluations Parent report of a diagnosis via questionnaire

DSM-IV-R criteria applied to developmental evaluations ADOS, maternal report, SRS

ADOS, SCQ

ASQ, BSID-I-II-III, DDSTII, MCDI, MIDI, MSCA,KBSID-II DSM-IV-R determined by DDS staff ADOS, SCQ

DSM-IV-R criteria applied to developmental evaluations A-TAC

ICM-9-CM

ASD measurement

Autistic disorder

Autistic disorder

ASD

ASD

ASD

ASD

ASD

ASD

National-Scale Air Toxics Assessment (NATA) 1990, 1996, 1999 and 2002 CALINE4 traffic based emissions model Interpolation between up to four regulatory monitors Distance to freeway

Spatiotemporal models using data from EPA Air Quality System and airport visibility National-Scale Air Toxics Assessment (NATA) 1996

Concentration from nearest regulatory monitor National-Scale Air Toxics Assessment (NATA) 2005 Multi-pollutant air monitoring campaign LUR models using manual forward step-wise linear regression, Space-time interpolation between regulatory monitors

LUR models based on 2008 and 2011 monitoring campaigns

ASD

Autistic disorder

Annual average from emission databases

Average concentration from nearest three monitoring stations National-Scale Air Toxics Assessment (NATA) 1996

Exposure measurement

ASD

ASD

ASD

ASD classification

Pregnancy trimesters and birth

Annual average for a year within two years of birth year Pregnancy trimesters and 1st year of life

2-

2+

2-

D

C

2+ 9 months preconception through 9 months of life Annual average for year close to birth year

2-

C

C

C

C

C

C

GR

2+

2+

2-

2+

2+

2+

2-

2+

LE

Pre-conception through 1st year of life

Pregnancy trimesters and entire pregnancy Pregnancy through 2nd year of life 3 months preconception, pregnancy, 1st and 2nd year of life

Pre-conception, pregnancy 1st and 9th years of life Pre-conception through birth

1–4 years preceding diagnosis Annual average for year close to birth year

Exposure timing

Abbreviations: ADI-R, Autism Diagnostic Interview-Revised; ADOS, Autism Diagnostic Observation Schedules; ASQ, Ages and Stages Questionnaire; BSID, Bayley Scales of Infant Development (I-first edition, II-second-edition, III-3rd Edition); CALINE4, California Line Source Dispersion Model; DDST II, Denver Development Screening Test II; DSM-IV-R, Diagnostic and Statistical Manual of Mental Disorders, 4th Edition, Text Revision; ICD-9_CM, International Classification of Diseases, 9th Revision, Clinical Modification; MCDI, McArthur Communicative Development Inventory; MIDI, Minnesota Infant Development Inventory; MSCA, McCarthy Scales of Children's Abilities; SCQ, Social Communication Questionnaire; SRS, Social Responsiveness Scale.

San Francisco Bay area, CA, USA

Raz et al., 2015

1994, 1996, 1998 and 2000

2005–2009

Pennsylvania, USA

217/224 216/4,971 211/219

Nested casecontrol Case-control

2005–2009

7594/75,635

Pennsylvania, USA

San Francisco Bay area, CA, USA and North Carolina, USA USA

Kalkbrenner et al., 2015

Becerra et al., 2013 Talbott et al., 2015b Talbott et al., 2015a

Birth cohort

9482

Not available (recruited between 1992–2008) 1995–2006

Guxens et al., 2015

383/2829

Retrospective cohort Nested casecontrol

The Netherlands, Germany, Italy, France Greece and Spain Los Angeles, CA, USA

Sweden

Gong et al., 2014

1992, 1994 and 1996

49,073

Study design

Twin study

West Virginia and North Carolina, USA

Kalkbrenner et al., 2010

1997–2000

Sample size

3426

China

Jung et al., 2013

Birth years

1992–2000

Location

Citation

Table 3 Studies on PM exposure and ASD.

M. Morales-Suárez-Varela et al.

Environmental Research 153 (2017) 150–160

Environmental Research 153 (2017) 150–160

M. Morales-Suárez-Varela et al.

Table 4 Summary of results in studies on PM2.5 exposure and ASD. Citation

Standardized ORs per 5 μg/ m3 increase

Adjusted ORs

OR

Lower 95% CI

Becerra et al., 2013 Mean [PM]=19.6 μg/m3 1.07 1 PM2.5 per 4.68 μg/m3 increase Guxens et al., 2015 Mean [PM]=8.4–22.4 μg/m3 PM2.5 per 5 μg/ 1.01 0.63 m3 increase Raz et al., 2015 Mean [PM]=14.6 μg/m3 1.57 1.22 PM2.5 per 4.42 μg/m3 increase 1st trimester 1.23 1.01 2nd trimester 1.27 1.05 3rd trimester 1.49 1.2 Talbott et al., 2015a Mean [PM]=14.5 μg/m3 PM2.5 per 2.84 μg/m3 increase Pre pregnancy 1.13 0.94 1st trimester 1.07 0.91 2nd trimester 1.04 0.88 3rd trimester 1.04 0.88 Pregnancy 1.2 0.88 Year 1 1.37 0.95 Year2 1.45 1.01 1.2 0.95 Pre pregnancy through 1st trimester 1.29 0.94 Pre pregnancy through 2nd trimester 1.46 0.98 Pre pregnancy through pregnancy 1.47 0.98 Pre pregnancy through year 1 1.51 1.01 Pre pregnancy through year 2 Volk et al., 2013 PM2.5 per 8.7 μg/m3 increase 1st trimester 1.22 0.96 2nd trimester 1.48 1.4 3rd trimester 1.4 1.11 Pregnancy 2.08 1.93 Year 1 2.12 1.45

Table 5 Summary of results in studies on PM10 exposure and ASD.

Upper 95% CI

OR

Lower 95% CI

Upper 95% CI

1.15

1.07

1

1.16

1.63

1.01

0.63

1.63

2.03

1.67

1.25

2.23

1.49 1.54 1.85

1.26 1.31 1.57

1.01 1.06 1.23

1.57 1.63 2.01

1.35 1.25 1.22 1.24 1.63 1.97 2.08 1.52

1.24 1.12 1.07 1.07 1.37 1.74 1.92 1.37

0.9 0.85 0.8 0.8 0.8 0.91 1.02 0.91

1.7 1.48 1.41 1.46 2.36 3.3 3.63 2.09

1.76

1.56

0.9

2.71

2.19

1.94

0.97

3.98

2.21

1.97

0.97

4.04

2.26

2.07

1.02

4.2

1.53 1.57 1.77 2.25 3.1

1.12 1.25 1.21 1.52 1.54

0.98 1.21 1.06 1.46 1.24

Citation

OR

Lower 95% CI

Becerra et al., 2013 Mean [PM]=36.3 μg/m3 1.03 0.96 PM10 per 8.25 μg/m3 increase Gong et al., 2014 Mean [PM]=3.3–4.2 μg/m3 Pregnancy 0.78 0.39 1st year life 0.92 0.47 Guxens et al., 2015 Mean [PM]=17–44 μg/m3 PM10 per 10 μg/ 0.92 0.55 m3 increase Jung et al., 2013 PM10 per 10 μg/m3 increase Preceeding 1 year 1.08 0.97 Preceeding 2 1.02 0.91 years Preceeding 3 0.94 0.84 years Preceeding 4 0.92 0.81 years 3 1.05 0.97 10 μg/m < 48.68 μg/m3 1 1 Reference 1.38 1.04 48.69–57.96 μg/ m3 1.36 0.98 57.97–70.57 μg/ m3 3 0.94 0.64 > 70.58 μg/m Kalkbrenner et al., 2015 Mean [PM]=22.9–25 μg/m3 PM10 per 10 μg/m3 increase Preconception 0.94 0.82 1st trimester 0.86 0.74 2nd trimester 0.97 0.83 3rd trimester 1.36 1.13 Post natal 1.09 0.9 quarter 1 Post natal 0.76 0.63 quarter 2 Post natal 0.85 0.7 quarter 3 Post natal 1.19 0.98 quarter 4 Raz et al., 2015 Mean [PM]=9.9 μg/m3 1.07 0.89 PM10–2.5 per 5.15 μg/m3 increase 1st trimester 1.05 0.92 2nd trimester 1.03 0.89 3rd trimester 1.07 0.92 Volk et al., 2013 PM10 per 14.6 μg/m3 increase 1st trimester 1.44 1.07 2nd trimester 1.83 1.35 3rd trimester 1.61 1.2 Pregnancy 2.17 1.49 Year 1 2.14 1.46

1.28 1.3 1.39 1.59 1.92

normalization was appropriate None of the articles pertaining to Diesel PM were appropriate for normalization and therefore no standardized ORs are presented (Table 6 and Fig. 4).

3.1.1. Study design and population This systematic review includes 4 cohort and 8 case control studies, drawn from 15 different study populations and 6 different countries. Sample sizes ranged from 443 (Talbott et al., 2015a) to 83,229 (Becerra et al., 2013) representing a total of 193,657 participants, not accounting for participant the overlap across publications utilizing data from the same study populations (Talbott et al., 2015a, 2015b; Kalkbrenner et al., 2010, 2015; Raz et al., 2015; Roberts et al., 2013). One study used a Taiwanese population (Jung et al., 2013). Within the studies that used European populations, one used Swedish, Dutch, Italian and Spanish populations (Guxens et al., 2015) and another that used an exclusively Swedish population (Gong et al., 2014). The

Standardized ORs per 5 μg/ m3 increase

Adjusted ORs

Upper 95% CI

OR

Lower 95% CI

Upper 95% CI

1.1

1.02

0.98

1.06

1.54

0.96

0.74

1.24

1.19 1.13

1.04 1.01

0.98 0.95

1.09 1.06

1.05

097

0.92

1.02

1.04

0.96

0.98

1.02

1.08 0.99 1.15 1.63 1.32

0.97 0.93 0.98 1.16 1.04

0.91 0.86 0.91 1.06 0.95

1.04 0.99 1.07 1.28 1.15

0.93

0.87

0.76

0.96

1.04

0.92

0.84

1.02

1.43

1.09

0.99

1.2

1.21 1.19 1.24

1.05 1.03 1.07

0.89 0.86 0.89

1.15 1.16 1.2

1.96 2.47 2.14 3.16 3.12

1.13 1.23 1.17 1.3 1.3

1.02 1.11 1.06 1.15 1.14

1.26 1.36 1.3 1.48 1.48

1.56 1.79

1.19 1 1.84 1.91 1.37

1.28

remaining 9 studies used American populations. Among the studies, 2 used study populations from the Nurses´ Health Study II (Raz et al., 2015; Roberts et al., 2013). Another 2 used a western Pennsylvania population (Talbott et al., 2015a, 2015b), one used a combination of North Carolina and West Virginia populations (Kalkbrenner et al., 2010), one used a Los Angeles population (Becerra et al., 2013), another used a combination of North Carolina and San Francisco

154

Environmental Research 153 (2017) 150–160

M. Morales-Suárez-Varela et al.

When the results are standardized, the association between ASD and PM2.5 more evident and clear. However, there seems to be no trend attributable to total PM concentration or exposure period. As a whole, the studies provide evidence of an association between PM2.5 exposure and ASD. This association is the highest among the three particulate types studied in this review and therefore PM2.5 appears to be the most harmful of the PM fractions. No definitive conclusions can be extracted from the results when it comes to the exposure period or PM level since there seems to be no correlation between them and the OR. This could lead us to the conclusion that the child is vulnerable to even low levels of PM during the whole period of gestation.

Table 6 Summary of results in studies on diesel PM exposure and ASD. Citation

OR

Lower 95% CI

Kalkbrenner et al., 2010 80th percentile vs 20th percentile (1.844 μg/m3 vs 1.025 μg/m3) Rural 0.9 0.5 Mixed 1 0.7 Urban 0.9 0.6 North Carolina 1.1 0.6 West Virginia 1 0.7 All 1.1 0.8 Roberts et al., 2013 2nd quintile 1.06 g/m3 1.1 0.6 3rd quintile 1.48 g/m3 1.3 0.7 3 4th quintile 2.00 g/m 1.2 0.6 3 2 1 5th quintile 4.40 g/m Talbott et al., 2015b 2nd quartile interview 0.291 μg/m3 1.15 0.66 2nd quartile birth certificate 1.38 0.93 0.296 μg/m3 1 0.57 3rd quartile interview 0.411 μg/m3 3rd quartile birth certificate 1.3 0.86 3 0.411 μg/m 3 1.04 0.59 4th quartile interview 0.660 μg/m 4th quartile birth certificate 1.25 0.83 0.614 μg/m3 Volk et al., 2011 < 309 m 1.86 1.03 < 309 m (decile) 2.48 1.17 < 309 m (residential histoy data) 2.22 1.16 309–647 m 0.96 0.58 647–1419 m 1.11 0.73 > 1419 m Reference 1 1 Windham et al., 2006 3.37 μg/m3 average Diesel PM 1.44 1.03

Upper 95% CI

1.6 1.6 1.5 1.9 1.5 1.5 2.2 2.5 2.5 4

3.2.2. PM10 Jung et al., 2013 found no association between PM10 exposure during the 4 years preceding birth and ASD diagnosis. Guxens et al., 2015 also found no clear evidence of a link between ASD and PM10, and PMcoarse exposure. Kalkbrenner et al., 2015, which studied PM10 exposure during pregnancy and the 1st year of life, shows data attributing a positive association between ASD diagnosis and exposure during the 3rd trimester. Becerra et al., 2013 presents results suggesting a slight increase in odds of ASD diagnosis per interquartile range (IQR) (8.25 μg/m3) increase in entire pregnancy exposure to PM10 in a single pollutant model. Raz et al., 2015 shows that exposure to PM10-2.5 (per 5.15 μg/m3 increase) during pregnancy has a positive association with diagnosis of ASD. Volk et al., 2013, as with PM2.5, shows the highest positive association between ASD diagnosis and PM10 exposure during pregnancy and more so during the 1st year of life of all the studies. When standardized, the ORs for PM10 are all inferior and with CIs, apart from Volk et al., 2013, that include the null association value. Therefore, the association is less clear for these particles than for PM2.5. As a whole, the studies provide evidence of a very weak association between PM10 exposure and ASD. The association for PM10 appears to be dependent on the total PM concentration more than anything, the IQR values are higher than those for PM2.5 even though the association seems weaker. The exposure period does not appear to factor into the association in this case. The association appears to be the weakest of the three here studied but is still of importance and cannot be ignored.

2 2.06 1.77 1.96 1.84 1.87

3.45 5.39 4.42 1.56 1.67 1

2.02

populations (Kalkbrenner et al., 2015) while another used exclusively a San Francisco population (Windham et al., 2006) and the last used a California population (Volk et al., 2013). The methodology used to measure PM concentrations and exposure differs among the studies included and is based primarily on the use of statistical models rather than direct measurements of the exposure of each subject which would be practically impossible to conduct. It must also be noted that the methods for measuring PM2.5 and PM10 are quite different to that used for diesel PM, the former are normally measured at street level while the latter is modeled at 10,000 feet. Another element that must be pointed out is the different exposure levels found in the different areas in which studies were conducted.

3.2.3. Diesel PM Kalkbrenner et al., 2010 shows no association with ASD for diesel PM exposure in either state or in any of the exposure areas. Gong et al. found no association between PM exposure during pregnancy or the 1st year of life and ASD. Talbott et al., 2015b shows a slightly positive association in the all quartiles of exposure however none are statistically significant. For the birth certificate unadjusted analyses, associations were of borderline significance for 2nd versus 1st quartile for diesel PM., Windham et al., 2006 presents such an association for diesel PM. Roberts et al., 2013 shows a positive and statistically significant association in the 5th quintile of diesel PM exposure particularly in boys [(OR) =2.3; 95% CI =1.1–4.9]. Volk et al., 2011 shows a positive association when residential proximity to a highway is < 309 m. As a whole, the studies provide evidence of an association between diesel PM exposure and ASD. For diesel PM, as happens with PM10 the most important contributing factor seems to be the total concentration. Higher levels of diesel PM reflect higher OR. However, we must note that a higher concentration of diesel PM is also associated with higher levels of confounders which could play an import role in this case. Given the prevalence of diesel PM and its associated co-pollutants, exposure to diesel PM poses a risk to neurodevelopment which must be addressed.

3.2. Particulate matter 3.2.1. PM2.5 Talbott et al., 2015a presents the adjusted ORs and unadjusted OR of ASD for a 2.84 μg/m3 IQR increase in PM2.5 from preconception to two years post birth. The OR are elevated for all individual time periods with significant OR for year 2 post birth. Guxens et al., 2015 found no clear evidence of a link between ASD and PM2.5 exposure. In this study, the data for PM2.5 is the most statistically significant and interestingly shows a negative association between PM exposure and ASD. Becerra et al., 2013 presents results suggesting a 5–15% relative increase in odds of ASD diagnosis per interquartile range (IQR) (4.68 μg/m3) increase in entire pregnancy exposure to PM2.5 in a single pollutant model. These estimates increase for the two- pollutant models including ozone and PM2.5. Raz et al., 2015 shows that exposure to PM2.5 during pregnancy, particularly the 3rd trimester, has a positive association with diagnosis of ASD. Volk et al., 2013 shows the highest positive association between ASD diagnosis and PM2.5 exposure during pregnancy and more so during the 1st year of life of all the studies.

4. Discussion We have arrived at conclusions similar to those on the four previous 155

Environmental Research 153 (2017) 150–160

M. Morales-Suárez-Varela et al.

Fig. 2. Standardized (per 5 μg/m³ increase) association between ASD and PM2.5.

based on diesel PM which as we have stated always has the possibility of the presence of confounders. When speaking specifically about PM2.5 or PM10 the studies on which it bases its conclusions are Windham et al., 2006, Kalkbrenner et al., 2010, Kalkbrenner et al., 2015, Volk et al., 2011, Volk et al., 2013, Becerra et al., 2013 and Roberts et al., 2013. All of which, with the exception of Kalkbrenner et al., 2010 which shows no association between diesel PM and ASD even when not accounting for confounders, have been mentioned previously above. Weisskopf et al., 2015 is a review of only two studies; Kalkbrenner et al., 2015 and Raz et al., 2015 which have already been discussed above. Of the four reviews, only Kalkbrenner et al., 2014 separate PM into categories as we have done in this review. This separation is crucial in order to probably ascertain the association between each of the PM fractions and ASD due to their different physical and chemical characteristics.

reviews, there exists an association between ASD and PM and the association is heavily dependent on size and PM composition. The findings of Rossignol et al., 2014 are based on Becerra et al., 2013, Windham et al., 2006, Roberts et al., 2013, Volk et al., 2011 and Volk et al., 2013. Windham et al., 2006, Volk et al., 2011 and Volk et al., 2013 show positive associations between ASD and PM exposure, however, we have given less weight to the studies by Volk due to their methods of measuring PM exposure (roadway emissions) and in Windham et al., 2006 the study is on diesel PM and there exist the possibility of the presence of confounding factors such as other traffic related pollutants. In Roberts et al., 2013 we once again have data for diesel PM with the corresponding possibility of confounders and in Becerra et al., 2013 the positive association between ASD and PM exposure appears in the two pollutant model which assumes the presence of at least one confounder. Suades-Gonzalez et al., 2014 bases its findings on the following studies; Becerra et al., 2013, Roberts et al., 2013, Kalkbrenner et al., 2015, Raz et al., 2015, Volk et al., 2013, Talbott et al., 2015a, Jung et al., 2013, Guxens et al., 2015 and Gong et al., 2014. The differences arising from the different interpretations of Becerra et al., 2013, Roberts et al., 2013 and Volk et al., 2013 have already been explained. Guxens et al., 2015 and Gong et al., 2014 arrived at the conclusion that there was no association between ASD and PM exposure which is more like our conclusion of the presence of a weak and limited link between ASD and PM exposure. The remaining studies; Kalkbrenner et al., 2015, Raz et al., 2015 and Talbott et al., 2015a are grouped together as showing positive association however in Kalkbrenner et al., 2015 the association appears only in the third trimester for PM10 per 10 μg/m3 increase, in Talbott et al., 2015a for PM2.5 per 2.84 μg/m3 increase only when year two post birth is included and in Raz et al., 2015 only when referring to PM2.5 per 4.42 μg/m3 increase. Taking into account the different exposure periods conditions in which the positive association appears in these three studies we could not confidently conclude that the association is due to PM exposure exclusively which is what we are analyzing in this review. Kalkbrenner et al., 2015 includes conclusions

4.1. Summary of evidence Our analysis presents summarized evidence of PM exposure and its associations with ASD in humans. While a risk of bias cannot be excluded, especially when in reference to accurate exposure assessment which could affect the ability to detect associations in epidemiological studies such as this one, in the articles reviewed there seems to be an association between PM exposure and the diagnosis of ASD heavily dependent on PM size and source. 4.2. Strengths and limitations of the current review It is possible that our search may not have retrieved all relevant publications relating to autism and particulate matter given that we restricted this review to studies published in English available through the PubMed database. This review, as any other review of observational data, may be sensitive to publication bias since it is believed that studies with significant positive results are more widely distributed 156

Environmental Research 153 (2017) 150–160

M. Morales-Suárez-Varela et al.

Fig. 3. Standardized (per 5 μg/m³ increase) association between ASD and PM10.

Volk et al., 2013 and Volk et al., 2011Comparing this group of studies to those that use other criteria for the diagnosis of ASD, we observe that the ADOS group presents higher associations between PM and ASD almost across the board.

than those without significant results or negative ones. 4.2.1. Conceptual constraints In the publications retrieved, the phrase “environmental pollution” carries a variety of meanings as there is no standardized definition of this concept. To focus our attention solely on pollutants of high global concern, in this review we chose to limit the definition of environmental pollution to particulate matter or diesel particulate matter. Therefore, all the literature included in this review presents data relating specifically to this pollutant. This proved to be a significant constraint since publications containing this specific information are very scarce as is evidenced by the low number of studies that were eligible for inclusion in this review.

4.3.2. Observation and exposure periods The observation period, which is a key factor when studying ASD (Lord et al., 2006), used in the studies reviewed is also not homogenous. If the studies are performed at too early of an age, the rate of detection or misdiagnosis could possibly be significant and alter the results. Along this line, we find that the exposure periods monitored are also different among the studies. Even though the studies mostly cover what is considered the period of susceptibility whereby xenobiotic agents such as PM may contribute to autism (pregnancy and the first year of life) there exists the possibility that in some studies parts of the critical period of exposure might have been excluded leading to misleading results (Volk et al., 2014).

4.3. Strengths and limitations of studies included in the review 4.3.1. ASD diagnosis The proper diagnosis of ASD is complex and the rates of detection and misdiagnosis may vary geographically due to differences in access to health and education services [Guxens et al., 2015]. The difference in diagnostic and inclusion criteria makes the comparison of results difficult and a possible source of bias. None of the studies use the criteria found in the Diagnostic and Statistical Manual of Mental Disorders, 5th Edition but four use DSM-IV. Of these four, (Kalkbrenner et al., 2010; Becerra et al., 2013; Kalkbrenner et al., 2015 and Windham et al., 2006) three (Becerra et al., 2013; Kalkbrenner et al., 2015 and Windham et al., 2006) show a positive association between ASD and PM exposure. Another common diagnostic criterion used is the Autism Diagnostic Observation Schedule (ADOS) (Talbott et al., 2015b; Talbott et al., 2015a; Raz et al., 2015;

4.3.3. PM characteristics As with the information on ASD diagnosis, the data relating to PM may also be a source of bias in this review. The most important characteristics relating to PM that could cause bias are the size and source of the particulate. This review has aimed to minimize this bias by separating PM into three groups of study. 4.3.4. PM exposure assessment The methodology used to measure PM concentrations and exposure differs among the studies included and is based primarily on the use of statistical models rather than direct measurements of the exposure of 157

Environmental Research 153 (2017) 150–160

M. Morales-Suárez-Varela et al.

Fig. 4. Association between ASD and Diesel PM.

integrate all the results in a single scale, this integration posed numerous problems given the nature of the included studies. The studies took place under very different conditions, specifically in areas with different total PM concentration levels and with different adjustment variables. ORs were standardized to the extent possible, given the data available and methodology utilized in the different studies. This was possible for only 6 of the 13 studies included (Tables 4 and 5 and Figs. 2 and 3).

each subject which would be practically impossible to conduct. The variability in these methodologies makes it difficult to group studies according to this characteristic. However, we can group together the studies that use data from the National-Scale Air Toxics Assessments (NATA) (Kalkbrenner et al., 2010; Talbott et al., 2015b; Windham et al., 2006; Roberts et al., 2013). While the source of the PM exposure data in these cases is the same, it must be noted that it is taken from different years. Three studies include information on diesel PM and are the ones that show slight positive associations between exposure and ASD (Talbott et al., 2015b; Windham et al., 2006; Roberts et al., 2013). Jung et al., 2013 and Becerra et al., 2013 are the only two studies that use information from the nearest monitoring stations and in these we observe that the 48.69–57.96 μg/m3 and 57.97–70.57 μg/m3 groups in the first study show a positive association as does the PM2.5 per 4.68μg/m3 increase group in the second.

4.3.7. Confounders Unaccounted confounding could also be a source of information bias in this review. While the reviewed studies accounted in some way for a number of potential confounding variables such as maternal age, maternal education, maternal smoking, sex, race and urbanicity not all variables were taken into account in all of the studies. While there seems to be no important differences between the crude and adjusted odds ratios in the studies reviewed with the exception of Kalkbrenner et al., 2015 and Jung et al., 2013, a standardized list of possible confounding variables for ASD, such as exposure to other pollutants, would help minimize any possible errors derived from unaccounted confounding. One important confounding variable less commonly taken into account is the presence of other pollutants that might affect the effects of PM. In several of the studies, the use of both one and multi pollutant models highlighted the possible association between PM and other pollutants (Jung et al., 2013; Becerra et al., 2013; Roberts et al., 2013). This association, if proven to be strong, could be an important confounding factor and further studies would be needed to correctly determine if the possible association between ASD and the pollutants is due to PM, the other pollutant(s) or the combination of both factors. In our review we have not taken into account the information from multi pollutant models since available data was scarce and the focus of this review is to study the association exclusively between PM and ASD.

4.3.5. Total PM concentration Another element to take into consideration is the vastly different total PM concentration values found in the different areas in which studies were conducted. Even in those articles in which OR are calculated on the basis of IQR, these ranges are not uniform which presents a challenge to the homogeneity of the data. This variability coupled with the previously mentioned difference in exposure assessment criteria could provide misleading information on the association between ASD diagnosis and PM exposure and therefore account for some of the contradictory findings. However, this variability in the data could also prove to be useful as it allows for the study of the possible association between ASD and PM exposure along a wide range of concentrations of the latter. This information could serve to provide a gradient of risk associated with the different values of PM concentrations. 4.3.6. Odds ratio calculations As mentioned previously, it was noticed that the ORs were calculated following different criteria, such as per increase in PM concentration, according to quartiles/quintiles, distance to freeways etc. Normally ORs should be standardized to the same unit to avoid possible confusion and all the results would best be presented in a single integrated scale. However; when the attempt was made to

4.4. Biological plausibility Oxidative stress may contribute to the pathogenesis of ASD through lipid peroxidation, reduced antioxidant activity, elevated nitric oxide level and inflammation (James et al., 2009; Chauhan et al., 2004; Ming 158

Environmental Research 153 (2017) 150–160

M. Morales-Suárez-Varela et al.

and interpretation of the results. Further, all have participated in the preparation of this manuscript and have approved the final version submitted for publication.

et al., 2005; Zoroğlu et al., 2004; Yorbik et al., 2002; Söğüt et al., 2003; Enstrom et al., 2009; Li et al., 2009). It is suspected that PM components are one of the major culprits of the neurological effects of air pollution (Block et al., 2012) since smaller PM particles may penetrate cellular membranes (Geiser et al., 2005; Rothen-Rutishauser et al., 2008) and translocate from the systemic circulation or via the nasal mucosa and the olfactory bulb to the lungs and into the brain. (Campbell et al., 2009; Oberdorster et al., 2009). In in vivo rodent studies, PM2.5 activates the stress axis and causes production of pro-inflammatory cytokines in the brain (MohanKumar et al., 2008) and may also alter the development of the neonatal immune system leading to a reduction in T cells and an increase in B lymphocytes in neonatal cord blood (Hertz-Picciotto et al., 2005). Neuroinflammation and early immune system activation are associated with ASD and are possible mechanisms by which environmental pollutants could increase the risk of ASD in humans (Atladóttir et al., 2010; Careaga et al., 2013; Depino et al., 2013; Gadad et al., 2013; Libbey et al., 2005; Patterson et al., 2011; Hertz-Picciotto et al., 2008). Several animal studies have indicated potential endocrine-disrupting effects, [Watanabe and Kurita, 2001] permanent alterations in both learning ability and activity of prenatal and neonatal exposure to diesel PM (USEPA, 2002) and increased brain inflammation indices in mice exposed to PM (Campbell et al., 2005). Exposure to environmental pollutants is determined by emissions sources which normally produce very complex mixtures. This makes it difficult to separate the effects of specific compound and on top of that, adverse health effects may be potentiated by joint exposures. A strong genetic component is indicated in the etiology of autism. It has been hypothesized that the presence of susceptibility genes, when combined with exposure to certain pollutants, will lead to this condition [London and Etzel, 2000]. If these findings are correct it could lead to the conclusion that in order to develop ASD the presence of a genetic predisposition for ASD alone is not enough and an exposure to pollutants during the critical period of development could act as a sort of trigger.

Funding The authors received no financial support for the research, authorship and/or publication of this article. Ethical approval This paper does not deal with research or studies on human subjects or experimental animals, and therefore no review and/or approval by an appropriately constituted committee for human subjects or animal research is needed. References Atladóttir, H.O., Thorsen, P., Østergaard, L., et al., 2010. Maternal infection requiring hospitalization during pregnancy and autism spectrum disorders. J. Autism Dev. Disord. 40, 1423–1430. Baxter, A., Brugha, T., Erskine, H., et al., 2015. The epidemiology and global burden of autism spectrum disorders. Psychol. Med. 45 (3), 601–613. Becerra, T.A., Wilhelm, M., Olsen, J., et al., 2013. Ambient air pollution and autismin Los Angeles County, California. Environ. Health Perspect. 121 (3), 380–386. Block, M.L., Elder, A., Auten, R.L., et al., 2012. The outdoor air pollution and brain health workshop. Neurotoxicology 33, 972–984. Campbell, A., Araujo, J.A., Li, H., et al., 2009. Particulate matter induced enhancement of inflammatory markers in the brains of apolipoprotein E knockout mice. J. Nanosci. Nanotechnol. 9 (8), 5099–5104. Campbell, A., Oldham, M., Becaria, A., et al., 2005. Particulate matter in polluted air may increase biomarkers of inflammation in mouse brain. Neurotoxicology 26 (1), 133–140. Careaga, M., Hansen, R.L., Hertz-Piccotto, I., et al., 2013. Increased anti-phospholipid antibodies in autism spectrum disorders. Mediat. Inflamm. 013, 935608. Chauhan, A., Chauhan, V., Brown, W.T., Cohen, I., 2004. Oxidative stress in autism: increased lipid peroxidation and reduced serum levels of ceruloplasmin and transferrin - the antioxidant proteins. Life Sci. 75, 2539–2549. Depino, A.M., 2013. Peripheral and central inflammation in autism spectrum disorders. Mol. Cell Neurosci. 53, 69–76. Elsabbagh, M., Divan, G., Koh, Y.-J., et al., 2012. Global prevalence of autism and other pervasive developmental disorders. Autism Res. 5 (3), 160–179. Enstrom, A., Krakowiak, P., Onore, C., et al., 2009. Increased IgG4 levels in children with autism disorder. Brain Behav. Immun. 23, 389–395. Friedenreich, C.M., 1993. Methods for pooled analyses of epidemiologic studies. Epidemiology 4, 295–302. Gadad, B.S., Hewitson, L., Young, K.A., German, D.C., 2013. Neuropathology and animal models of autism: genetic and environmental factors. Autism Res. Treat.. Geiser, M., Rothen-Rutishauser, B., Kapp, N., et al., 2005. Ultrafine particles cross cellular membranes by nonphagocytic mechanismsin lungs and in cultured cells. Environ. Health Perspect. 113 (11), 1555–1560. Geschwind, D., Levitt, P., 2007. Autism spectrum disorders: developmental disconnection syndromes. Curr. Opin. Neurobiol. 17 (1), 103–111. Gong, T., Almqvist, C., Bölte, S., et al., 2014. Exposure to air pollution From traffic and neurodevelopmental disorders in Swedish twins. Twin Res. Hum. Genet. 17, 553–562. Grandjean, P., Landrigan, P., 2014. Neurobehavioural effects of developmental toxicity. Lancet Neurol. 13 (3), 330–338. Guxens, M., Ghassabian, A., Gong, T., et al., 2015. Air pollution exposure during pregnancy and childhood autistic traits in four european. Environ. Health Perspect. 124, 133–140. Hallmayer, J., Cleveland, S., Torres, A., et al., 2011. Genetic heritability and shared environmental factors among twin pairs with autism. Arch. Gen. Psychiatry 68 (11), 1095–1102. Hertz-Picciotto, I., Herr, C.E.W., Yap, P.S., et al., 2005. Air pollution and lymphocyte phenotype proportions in cord blood. Environ. Health Perspect. 113, 1391–1398. Hertz-Picciotto, I., Park, H.Y., Dostal, M., et al., 2008. Prenatal exposures to persistent and non-persistent organic compounds and effects on immune system development. Basic Clin. Pharmcol. Toxicol. 102, 146–154. James, S.J., Rose, S., Melnyk, S., et al., 2009. Cellular and mitochondrial glutathione redox imbalance in lymphoblastoid cells derived from children with autism. FASEB J. 23, 2374–2383. Jung, C., Lin, Y., Hwang, B., 2013. Air pollution and newly diagnostic autism spectrum disorders: a population-based cohort study in Taiwan. PLoS One 8 (9), e75510. Kalkbrenner, A.E., Daniels, J.L., Chen, J.C., et al., 2010. Perinatal exposure to hazardous air pollutants and autism spectrum disorders at age 8. Epidemiology 21 (5), 631–641. Kalkbrenner, A.E., Schmidt, R.J., Penlesky, A.C., 2014. Environmental chemical exposures and autism spectrum disorders: a review of the epidemiological evidence. Curr. Probl. Pediatr. Adolesc. Health Care 44 (10), 277–318.

5. Recommendations and conclusion Given the results, we cannot state that there is no association between PM exposure and diagnosis of ASD. On the basis of this review, we conclude that the evidence to support a link between PM exposure and ASD is heavily dependent on size. Data today is still insufficient to reach a clear consensus on whether PM exposure significantly raises the incidence of ASD diagnosis. Thus, based on the results of this review, additional data is needed to ascertain the nature of the relationship between PM exposure and ASD. Studies with carefully collected data on PM exposure and other potential confounders and using the standardized ASD diagnosis criteria of DSM-V, are needed and will likely be useful in filling the gap that exist in the literature on this seemingly complex relationship. Future research is required and should be focused on PM exposure during late pregnancy and the first years of life. One pollutant models are necessary to establish the possible effects of PM exposure but should always be accompanied by multi pollutant models given that PM concentrations are highly correlated to other air pollutants specially those derived from traffic emissions. Competing interests The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article. Authors’ contributions All three coauthors of this paper have contributed significantly to the design and implementation of the review, as well as the analysis 159

Environmental Research 153 (2017) 150–160

M. Morales-Suárez-Varela et al.

nanoparticles. ALTEX 25 (3), 191–196. Sackett, D.L., 1997. Evidence-based medicine. Semin. Perinatol. 21 (1), 3–5. Sandin, S., Lichtenstein, P., Kuja-Halkola, R., et al., 2014. The familial risk of autism. JAMA 311 (17), 1770–1777. Scottish Intercollegiate Guidelines Network, 2008. A guideline developers' handbook (Publication no. 50). Edinburgh: SIGN: 2001 (updated in January. Söğüt, S., Zoroğlu, S.S., Özyurt, H., et al., 2003. Changes in nitric oxide levels and antioxidant enzyme activities may have a role in the pathophysiological mechanisms involved in autism. Clin. Chim. Acta 331 (1), 111–117. Suades-González, E., Gascon, M., Guxens, M., Sunyer, J., 2015. Air pollution and neuropsychological development: a review of the latest evidence. Endocrinology 156 (10), 3473–3482. Talbott, E.O., Arena, V.C., Rager, J.R., et al., 2015a. Fine particulate matter and the risk of autism spectrum disorder. Environ. Res. 140, 414–420. Talbott, E.O., Marshall, L.P., Rager, J.R., et al., 2015b. Air toxics and the risk of autism spectrum disorder: the results of a population based case–control study in southwestern Pennsylvania. Environ. Health 14, 80. U.S. EPA, 2002. Office of research and development. Health assessment document for diesel engine exhaust (EPA/600/8-90/057F). U.S. Environmental Protection Agency, Washington, DC. Volk, H.E., Lurmann, F., Penfold, B., et al., 2013. Traffic-related air pollution, particulate matter, and autism. JAMA Psychiatry 70 (1), 71–77. Volk, H.E., Kerin, T., Lurmann, F., et al., 2014. Autism spectrum disorder: interaction of air pollution with the MET receptor tyrosine kinase gene. Epidemiol. Camb. Mass 25 (1), 44–47. Watanabe, N., Kurita, M., 2001. The masculinization of the fetus during pregnancy due to inhalation of diesel exhaust. Environ. Health Perspect. 109, 111–119. Weisskopf, M.G., Kioumourtzoglou, M.A., Roberts, A.L., 2015. Air pollution and autism spectrum disorders: causal or confounded? Curr. Environ. Health Rep. 2 (4), 430–439. Windham, G.C., Zhang, L., Gunier, R., et al., 2006. Autism spectrum disorders in relation to distribution of hazardous air pollutants in the san francisco bay area. Environ. Health Perspect. 114 (9), 1438–1444. World Health Organization, 2006. Air quality guidelines: Global update 2005. World Health Organization, 2013. Review of Evidence on Health Aspects of Air Pollution e REVIHAAP Project. Copenhagen, Denmark. Yorbik, O., Sayal, A., Akay, C., et al., 2002. Investigation of antioxidant enzymes in children with autistic disorder. Prostaglandins Leukot. Ess. Fat. Acids 67, 341–343. Zoroğlu, S.S., Armutcu, F., Ozen, S., et al., 2004. Increased oxidative stress and altered activities of erythrocyte free radical scavenging enzymes in autism. Eur. Arch. Psychiatry Clin. Neurosci. 254, 143–147.

Kalkbrenner, A.E., Windham, G.C., Serre, M.L., et al., 2015. particulate matter exposure, prenatal and postnatal windows of susceptibility, and autism spectrum disorders. Epidemiology. 26 (1), 30–42. Lai, M.-C., Lombardo, M., Baron-Cohen, S., 2014. Autism. Lancet 383, 896–910. Lawler, C.P., 2008. The ‘environment’ for autism research: signs of improvement? Environ. Health Perspect. 116, A416–A417. Li, X., Chauhan, A., Sheikh, A.M., et al., 2009. Elevated immune response in the brain of autistic patients. J. Neuroimmunol. 207, 111–116. Libbey, J.E., Sweeten, T.L., McMahon, W.M., Fujinami, R.S., 2005. Autistic disorder and viral infections. J. Neurovirol 11, 1–10. Liberati, A., Altman, D.G., Tetzlaff, J., et al., 2009. The PRISMA statement for reporting systematic reviews and meta-analyses of studies that evaluate healthcare interventions: explanation and elaboration. BMJ 339, b2700. London, E., Etzel, R.A., 2000. The environment as an etiologic factor in autism: a new direction for research. Environ. Health Perspect. 108 (suppl 3), 401–404. Lord, C., Risi, S., DiLavore, P.S., et al., 2006. Autism from 2 to 9 years of age. Arch. Gen. Psychiatry 63 (6), 694–701. Lyall, K., Schmidt, R.J., Hertz-Picciotto, I., 2014. Maternal lifestyle and environmental risk factors for autism spectrum disorders. Int J. Epidemiol. 43, 443–464. Ming, X., Stein, T.P., Brimacombe, M., et al., 2005. Increased excretion of a lipid peroxidation biomarker in autism. Prostaglandins Leukot. Ess. Fat. Acids 73, 379–384. MohanKumar, S.M.J., Campbell, A., Block, M., Veronesi, B., 2008. Particulate matter, oxidative stress and neurotoxicity. Neurotoxicology 29, 479–488. Newschaffer, C.J., Croen, L.A., Daniels, J., et al., 2007. The epidemiology of autism spectrum disorders. Annu Rev. Public Health 28, 235–258. Oberdorster, G., Elder, A., Rinderknecht, A., 2009. Nanoparticles and the brain: cause for concern? J. Nanosci. Nanotechnol. 9 (8), 4996–5007. Patterson, P.H., 2011. Maternal infection and immune involvement in autism. Trends Mol. Med 17, 389–394. Raz, R., Roberts, A.L., Lyall, K., et al., 2015. Autism spectrum disorder and particulate matter air pollution before, during, and after pregnancy: a nested case–control analysis within the nurses' health study II Cohort. Environ. Health Perspect. 123 (3), 264–270. Roberts, A.L., Lyall, K., Hart, J.E., et al., 2013. Perinatal air pollutant exposure and autism spectrum disorder in the children of the nurses' health study II. Environ. Health Perspect. 121, 978–984. Rossignol, D.A., Genuis, S.J., Frye, R.E., 2014. Environmental toxicants and autism spectrum disorders: a systematic review. Transl. Psychiatry 4 (2), e360. Rothen-Rutishauser, B., Mueller, L., Blank, F., et al., 2008. A newly developed in vitro model of the human epithelial airway barrier to study the toxic potential of

160