Original Research

RMMJ Rambam Maimonides Medical Journal Rambam Health Care Campus 2019 October; 10(4): e0021. ISSN: 2076-9172
Published online 2019 October 29. doi: 10.5041/RMMJ.10375.

Biomarkers for Autism Spectrum Disorders (ASD): A Meta-analysis

Ashley Ansel, M.Sc.,1 Yehudit Posen, Ph.D.,1,2 Ronald Ellis, Ph.D.,1,3 Lisa Deutsch, Ph.D.,4 Philip D. Zisman, Ph.D.,1 and Benjamin Gesundheit, M.D., Ph.D.1*

1Cell-El Therapeutics Ltd, Jerusalem, Israel
2PSW Ltd, Rehovot, Israel
3Biotech & Biopharma Consulting, Jerusalem, Israel
4Biostats Statistical Consulting Ltd, Modiin, Israel

*To whom correspondence should be addressed. E-mail: GesundheitMD@cell-el.com

Abstract

Objective
To compare the reported accuracy and sensitivity of the various modalities used to diagnose autism spectrum disorders (ASD) in efforts to help focus further biomarker research on the most promising methods for early diagnosis.

Methods
The Medline scientific literature database was searched to identify publications assessing potential clinical ASD biomarkers. Reports were categorized by the modality used to assess the putative markers, including protein, genetic, metabolic, or objective imaging methods. The reported sensitivity, specificity, area under the curve, and overall agreement were summarized and analyzed to determine weighted averages for each diagnostic modality. Heterogeneity was measured using the I2 test.

Results
Of the 71 papers included in this analysis, each belonging to one of five modalities, protein-based followed by metabolite-based markers provided the highest diagnostic accuracy, each with a pooled overall agreement of 83.3% and respective weighted area under the curve (AUC) of 89.5% and 88.3%. Sensitivity provided by protein markers was highest (85.5%), while metabolic (85.9%) and protein markers (84.7%) had the highest specificity. Other modalities showed degrees of sensitivity, specificity, and overall agreements in the range of 73%–80%.

Conclusions
Each modality provided for diagnostic accuracy and specificity similar or slightly higher than those reported for the gold-standard Autism Diagnostic Observation Schedule (ADOS) instrument. Further studies are required to identify the most predictive markers within each modality and to evaluate biological pathways or clustering with possible etiological relevance. Analyses will also be necessary to determine the potential of these novel biomarkers in diagnosing pediatric patients, thereby enabling early intervention.

Keywords: Autism spectrum disorder, biomarkers, gene expression, magnetic resonance imaging, meta-analysis, proteomics

INTRODUCTION

Autism spectrum disorders (ASD) were first characterized clinically in 1943 by Kanner1 and further in 1979 by Wing and Gould2 as a spectrum of impaired social interactions, restricted communications skills, and unusual repetitive behaviors. The American Psychological Association’s Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5) recently consolidated the various subtypes of pervasive developmental disorders (PDD) into one category called ASD and shifted the evaluation from three domains (social deficits, communication deficits, and restricted repetitive behaviors [RRB]) to two (social-communication impairments and RRB).3

Several behavior assessment-based diagnostic tests have been used for ASD, including the Autism Diagnostic Observation Schedule (ADOS).4 A meta-analysis involving ADOS evaluations of >4,000 children reported an overall diagnostic accuracy of 52%, with sensitivity scores of 67%–97% (pooled data: 91%) and specificity scores of 56%–94% (pooled data: 73%).5 The Autism Diagnostic Interview–Revised (ADI-R) is a structured interview of the parent, differing from the direct observation of the child performed with the ADOS evaluation.6 In addition, clinicians use the Childhood Autism Rating Scale (CARS) to rate the child’s behavior on 15 subscales;7 parental reports can also be considered.

Since these behavioral tests are subjective and time-consuming, require professional staff to be administered, and can only be used from age 3 years once the child is old enough to communicate, researchers have sought other ways of diagnosing ASD.8 Biomarkers are expected to be more objective, should enable earlier diagnosis, and may provide clues to the underlying etiology of ASD. In addition, providing positive diagnosis in younger toddlers may enable earlier initiation of therapy with consequently higher probability of successful treatment given decreasing brain plasticity with age in the developing child. The primary modalities harnessed to identify novel ASD biomarkers have been molecular, proteomic, metabolomic, neurochemical, radiologic, and electrophysiologic, with transcriptomic analyses also having been performed.9

This study aimed to compare the reported accuracy and sensitivity of the various modalities used to diagnose ASD. This should help focus further biomarker research on the most promising methods for early diagnosis.

METHODS

This analysis aimed to adhere to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines, to ensure comprehensive and transparent reporting.

Search Strategy
Independent queries using medical subject headings (MeSH) and keywords were performed to identify all primary research articles from the PubMed database that evaluated sensitivity, specificity, and accuracy of biomarkers for diagnosing ASD within six predefined modalities (protein, metabolic, genetic, electroencephalography [EEG], magnetic resonance imaging [MRI], positron emission tomography [PET]) (Supplement). Each search term included autism/ASD, a diagnostic technique, and a sensitivity/specificity classifier. To distinguish between protein and metabolic papers, all papers dealing with hormones, urine, mass spectrophotometry, metabolites, or peptides were assigned to the metabolic modality, while all papers dealing with cytokines, chemokines, or other proteins circulating in the blood were assigned to the protein modality.

Data Sources and Data Extraction
The search terms were entered into Ovid MEDLINE (1946 to January 2017) without limits, and 866 articles were returned. Reviews and reference lists were cross-checked for studies that the search terms might have missed.

Screening. Two independent reviewers (AA and JR) examined study titles. From review of the abstracts, potentially eligible full-text articles were retrieved with relevant appendices and supplementary information.

Eligibility. Full-text articles were reviewed against eligibility criteria. Inclusion criteria were: (a) articles after 1994, written in English; (b) inclusion of a typically developing (TD) control group (unless it was a review paper); (c) inclusion of a study group with children diagnosed with ASD by a behavioral diagnostic test or by DSM criteria (unless it was a review paper); and (d) assessment for ASD biomarkers using one of the six predefined modalities. Exclusion criteria were: (a) studies that compared ASD with other comorbidities; (b) studies that tested only risk factors of ASD; (c) studies that had a therapeutic component; (d) non-clinical studies; and (e) studies without statistical parameters of interest (sensitivity, specificity, accuracy, and/or area under the curve [AUC]). All publications meeting all inclusion criteria and none of the exclusion criteria were included in the analysis.

Data Extraction. For each eligible article, the following data were extracted and validated independently by two researchers (AA and JR): first author’s surname, year of publication, diagnostic modality, number of ASD subjects, number of controls, age-matched (yes or no), sex-matched (yes or no), accuracy (% correctly identified), AUC, sensitivity, and specificity. The standard error of the AUC was calculated based on the AUC point estimate and sample size by the method of Hanley and McNeil.10 Several papers included the evaluation of multiple markers assessed on the same group of subjects; these were included in the meta-analysis individually and analyzed independently of each other.

Data Extraction. For each eligible article, the following data were extracted and validated independently by two researchers (AA and JR): first author’s surname, year of publication, diagnostic modality, number of ASD subjects, number of controls, age-matched (yes or no), sex-matched (yes or no), accuracy (% correctly identified), AUC, sensitivity, and specificity. The standard error of the AUC was calculated based on the AUC point estimate and sample size by the method of Hanley and McNeil.10 Several papers included the evaluation of multiple markers assessed on the same group of subjects; these were included in the meta-analysis individually and analyzed independently of each other.

Data Analysis. A separate statistical analysis of each of the parameters was performed for each of the six predefined modalities. Meta-analyses of sensitivity, specificity, AUC, and accuracy were performed using MedCalc Statistical Software version 17.9.7 (MedCalc Software bvba, Ostend, Belgium). The weighted summary sensitivity, specificity, accuracy (arc-sine square root transformation), and AUC with 95% confidence intervals (CIs) were calculated using a random-effects model.11 Inter-study heterogeneity was assessed using the I2 statistic with 95% CI, which describes the percent variability in point estimate due to heterogeneity rather than sampling error. Presented are the weighted summaries of each tested parameter, alongside the weighted summary of the ADOS test, as reported in a published meta-analysis.5 The contribution of each individual publication to the weighted summary of each measured parameter is presented by modality.

RESULTS

The literature search identified 866 papers (Figure 1), of which 211 were duplicates. The abstracts of the remaining 655 papers were screened for relevance; 29 were excluded. Thus, 626 full-text papers were assessed for eligibility based on the above criteria. Review of these papers identified 86 additional papers which were also further appraised. Of these 712 publications in total, 640 failed to meet the inclusion criteria, primarily due to lack of an ASD group or insufficient statistical data. The 72 papers that met the criteria were subdivided by the diagnostic modality reported in the study. Four papers included data belonging to two diagnostic modalities.1215 Since only one PET study met the eligibility criteria, this modality was excluded from the analysis. Cohort sizes varied from 6 to 554 subjects.

Figure 1Figure 1
PRISMA Flow Diagram of the Phases of the Literature Search.

Genetic studies (n=14)13,14,1627 applied polymerase chain reaction (PCR) genotyping, mRNA/miRNA microarray, or spectrophotometric tools. Magnetic resonance imaging (MRI) studies (n=22),13,14,21,2846 which included functional MRI (fMRI), resting state fMRI (rs-fMRI), structural MRI (sMRI), and standard MRI studies, sought out ASD diagnostic markers by analyzing fast spin echo (FSE) T2-weighted, fluid-attenuated inversion recovery (FLAIR), diffusion-weighted imaging (DWI), spin echo (SE) T1-weighted sequences, and single voxel 1H MR spectrum. Volumetric measurements for different areas of the brain were assessed. Mass univariate methods such as voxel-based morphometry (VBM) and whole-brain classification approach employing a support vector machine (SVM) were used. In the 21 metabolic studies,12,15,4765 high-performance liquid chromatography (HPLC), liquid chromatography–high resolution mass spectrometry (LC-HRMS), gas chromatography, nuclear magnetic resonance (NMR), as well as capillary electrophoresis with ultraviolet/ visible spectroscopy (UV-Vis) were used to identify markers. The 12 studies assessing the diagnostic potential of various protein markers12,15,6675 covered >60 proteins or protein combinations using various microarray kits/chips or enzyme-linked immunosorbent assay (ELISA). Six studies focused on EEG analyses for ASD diagnosis.7681

Overall, protein-based followed by metabolite-based studies provided the highest diagnostic accuracy, each with an overall agreement of 83.3% and AUC of sensitivity (true-positive rates) versus specificity (false-positive rates) of 89.5% and 88.3%, respectively (Table 1, Figures 25). Sensitivity provided by protein markers (85.5%) and metabolite markers (84.7%) was highest. The other modalities showed similar degrees of sensitivity, specificity, and overall agreements, which all fell within the range of 73%–80%.

Table 1Table 1
Weighted ASD Diagnostic Power for Each Evaluated Modality.
Figure 2Figure 2
Weighted Sensitivity of Appraised Studies.
Figure 3Figure 3
Weighted Specificity of Appraised Studies.
Figure 4Figure 4
Weighted Overall Agreement of Appraised Studies.
Figure 5Figure 5
Weighted AUC of Appraised Studies.

DISCUSSION

As the first reported meta-analysis of ASD biomarkers, the current study included 71 papers with cohorts of up to 554 ASD subjects.77 This study suggests that all five major bio-diagnostic modalities provide similar diagnostic objective accuracy for ASD compared to the subjective ADOS. The protein- and metabolite-based tests were found to provide for the highest diagnostic accuracy; combining modalities might further improve diagnostic accuracy.

The sensitivities of the various studied modalities were 73.6%–85.5%, with protein markers showing the highest degree of sensitivity. The specificities ranged from 73.0% (MRI) to 85.9% (metabolic). Accuracy, assessed by overall agreement and AUC, was 73.4%–83.2% and 79.0%–89.5%, respectively, and highest for protein-based biomarkers. Taken together, all analyzed modalities provided for higher diagnostic accuracy and specificity compared to the gold-standard ADOS test.5 While pooled ADOSdiagnostic sensitivity was higher than for biomarker modalities, protein-based diagnoses provided for sensitivity within the same range. This meta-analysis supports efforts to search for/use new objective modalities beyond psychological tests for the diagnosis of ASD. Moreover, quantitative objective biomarkers identified at ages when psychological tests cannot yet be employed should enable earlier-stage intervention, which is projected to be more efficient due to greater brain plasticity. These diagnostic efforts may enable the subdivision of ASD into subgroups and provide useful therapeutic targets, which have significant long-term therapeutic implications.

Further studies will be necessary to determine which modalities serve better as screening versus confirmatory testing. Subgroups of ASD might be defined by these tests, suggesting different therapeutic modalities as diagnostic targets for different such subgroups. Attempts to find patterns linking the most accurate biomarkers in each modality may identify common pathways and draw the ASD community closer to developing therapeutics, where these diagnostic markers will serve with psychological tests as objective theragnostic monitoring tools.

There were several limitations in this meta-analysis. There was a high level of heterogeneity, as expected given both the clinical heterogeneity between the included papers, with variations in the diagnosis and definition of ASD, and the methodologies across studies. In addition, in all assessed publications, evaluation of the diagnostic value of the biomarker of interest used typically developing controls as a comparator group, which may falsely elevate the diagnostic capacity of the test as compared to its performance in marginal cases with behaviors consistent with ASD. Comparing across modalities was not uniformly well-controlled. In addition, certain modalities were documented in a very limited number of papers. Furthermore, many papers were excluded due to insufficient statistical data, reinforcing the importance of proper study design and execution in future analyses of ASD biomarkers.

In conclusion, this study suggests that five major bio-diagnostic modalities provide a higher level of accuracy for objective diagnosis of ASD compared to ADOS, the gold-standard test. Thus, it is justified to include objective biological tests in the diagnosis of ASD to develop and monitor future biological therapies. Further studies looking at each modality in higher resolution to fine-tune the findings are still necessary. Objective biomarkers together with current psychological evaluations might enable improved diagnosis and monitoring.

Supplementary Data

Acknowledgements

The authors thank Joshua Rosenzweig for reviewing the literature, Frederic and Netanel Deutsch for assisting in the statistical analysis. The study was supported by Cell-El Ltd, a company seeking to define ASD-specific biomarker signatures to enable objective, biomarker-based ASD diagnostics.

Abbreviations

ADI-R Autism Diagnostic Interview–Revised
ADOS Autism Diagnostic Observation Schedule
ASD autism spectrum disorders
AUC area under the curve
CARS Childhood Autism Rating Scale
DSM Diagnostic and Statistical Manual of Mental Disorders
EEG electroencephalography
ELISA enzyme-linked immunosorbent assay
fMRI functional magnetic resonance imaging
HPLC high-performance liquid chromatography
LC-HRMS liquid chromatography–high resolution mass spectrometry
MRI magnetic resonance imaging
NMR nuclear magnetic resonance
PDD pervasive developmental disorders
PCR polymerase chain reaction
PET positron emission tomography
PRISMA Preferred Reporting Items for Systematic Reviews and Meta-Analyses
RRB restricted repetitive behaviors
rs-fMRI resting state functional magnetic resonance imaging
sMRI structural magnetic resonance imaging
SVM support vector machine
TD typically developing
UV-Vis ultraviolet/visible spectroscopy
VBM voxel-based morphometry.

Footnotes

Author contributions statement: AA reviewed the relevant literature, made substantial contributions to conception and design, and interpreted the data. YP reviewed the relevant literature and assisted in drafting the article. RE interpreted the findings and participated in revising the article and approving the final version. LD performed the statistical analyses. PDZ made substantial contributions to conception and design and participated in interpreting the data and in revising the paper. BG conceived the performing of the meta-analysis, developed its original detailed plan, participated in revising the paper, and approved the final version. All authors approved the final manuscript as submitted and agree to be accountable for all aspects of the work.

Conflict of interest: AA, PDZ, and BG are employees of Cell-El Ltd, a company seeking to identify autism-specific biomarkers. YP, RE, and LD serve as consultants at Cell-El Ltd. Cell-El Ltd had no control over the study data, analysis, interpretation, or the decision to publish.

REFERENCES
1.
Kanner L. Autistic disturbances of affective contact. Acta Paedopsychiatr. 1968;35:100–36.
2.
Wing L, Gould J. Severe impairments of social interaction and associated abnormalities in children: epidemiology and classification. J Autism Dev Disord. 1979;9:11–29.
3.
Lord C, Bishop SL. Recent advances in autism research as reflected in DSM-5 criteria for autism spectrum disorder. Annu Rev Clin Psychol. 2015;11:53–70. https://doi.org/10.1146/annurev-clinpsy-032814-112745.
4.
Lord C, Rutter M, Goode S, et al. Autism diagnostic observation schedule: a standardized observation of communicative and social behavior. J Autism Dev Disord. 1989;19:185–212.
5.
Tsheringla S, Minju KA, Russell S, Mammen P, Russell PS, Nair MK. A meta-analysis of the diagnostic accuracy of Autism Diagnostic Observation Schedule Module-1 for autism spectrum disorders. Indian J Pediatr. 2014;81(Suppl 2):S187–92. https://doi.org/10.1007/s12098-014-1627-9.
6.
Lord C, Rutter M, Le Couteur A. Autism Diagnostic Interview-Revised: a revised version of a diagnostic interview for caregivers of individuals with possible pervasive developmental disorders. J Autism Dev Disord. 1994;24:659–85.
7.
Schopler E, Reichler RJ, DeVellis RF, Daly K. Toward objective classification of childhood autism: Childhood Autism Rating Scale (CARS). J Autism Dev Disord. 1980;10:91–103.
8.
Gesundheit B, Rosenzweig JP. Editorial: Autism spectrum disorders (ASD)—searching for the biological basis for behavioral symptoms and new therapeutic targets. Front Neurosci. 2017;10:607. https://doi.org/10.3389/fnins.2016.00607.
9.
Ansel A, Rosenzweig JP, Zisman PD, Melamed M, Gesundheit B. Variation in gene expression in autism spectrum disorders: an extensive review of transcriptomic studies. Front Neurosci. 2017;10:601. https://doi.org/10.3389/fnins.2016.00601.
10.
Hanley JA, McNeil BJ. The meaning and use of the area under a receiver operating characteristic (ROC) curve. Radiology. 1982;143:29–36. https://doi.org/10.1148/radiology.143.1.7063747.
11.
DerSimonian R, Laird N. Meta-analysis in clinical trials. Control Clin Trials. 1986;7:177–88.
12.
El-Ansary A, Al-Ayadhi L. Neuroinflammation in autism spectrum disorders. J Neuroinflammation. 2012;9:265. https://doi.org/10.1186/1742-2094-9-265.
13.
Iidaka T. Resting state functional magnetic resonance imaging and neural network classified autism and control. Cortex. 2015;63:55–67. https://doi.org/10.1016/j.cortex.2014.08.011.
14.
Kang DW, Park JG, Ilhan ZE, et al. Reduced incidence of Prevotella and other fermenters in intestinal microflora of autistic children. PLoS One. 2013;8:e68322. https://doi.org/10.1371/journal.pone.0068322.
15.
Yang CJ, Liu CL, Sang B, Zhu XM, Du YJ. The combined role of serotonin and interleukin-6 as biomarker for autism. Neuroscience. 2015;284:290–6. https://doi.org/10.1016/j.neuroscience.2014.10.011.
16.
Croen LA, Goines P, Braunschweig D, et al. Brain-derived neurotrophic factor and autism: maternal and infant peripheral blood levels in the Early Markers for Autism (EMA) Study. Autism Res. 2008;1:130–7. https://doi.org/10.1002/aur.14.
17.
Gaita L, Manzi B, Sacco R, et al. Decreased serum arylesterase activity in autism spectrum disorders. Psychiatry Res. 2010;180:105–13. https://doi.org/10.1016/j.psychres.2010.04.010.
18.
Glatt SJ, Tsuang MT, Winn M, et al. Blood-based gene expression signatures of infants and toddlers with autism. J Am Acad Child Adolesc Psychiatry. 2012;51:934–44.e2. https://doi.org/10.1016/j.jaac.2012.07.007.
19.
Guan J, Yang E, Yang J, Zeng Y, Ji G, Cai JJ. Exploiting aberrant mRNA expression in autism for gene discovery and diagnosis. Hum Genet. 2016;135:797–811. https://doi.org/10.1007/s00439-016-1673-7.
20.
Hicks SD, Ignacio C, Gentile K, Middleton FA. Salivary miRNA profiles identify children with autism spectrum disorder, correlate with adaptive behavior, and implicate ASD candidate genes involved in neurodevelopment. BMC Pediatr. 2016;16:52. https://doi.org/10.1186/s12887-016-0586-x.
21.
Katuwal GJ, Cahill ND, Baum SA, Michael AM. The predictive power of structural MRI in autism diagnosis. Conf Proc IEEE Eng Med Biol Soc. 2015;2015:4270–3. https://doi.org/10.1109/EMBC.2015.7319338.
22.
Latkowski T, Osowski S. Computerized system for recognition of autism on the basis of gene expression microarray data. Comput Biol Med. 2015;56:82–8. https://doi.org/10.1016/j.compbiomed.2014.11.004.
23.
Maekawa M, Yamada K, Toyoshima M, et al. Utility of scalp hair follicles as a novel source of biomarker genes for psychiatric illnesses. Biol Psychiatry. 2015;78:116–25. https://doi.org/10.1016/j.biopsych.2014.07.025.
24.
Meng WD, Sun SJ, Yang J, Chu RX, Tu W, Liu Q. Elevated serum brain-derived neurotrophic factor (BDNF) but not BDNF gene Val66Met polymorphism is associated with autism spectrum disorders. Mol Neurobiol. 2016;54:1167–72. https://doi.org/10.1007/s12035-016-9721-9.
25.
Mundalil Vasu M, Anitha A, Thanseem I, et al. Serum microRNA profiles in children with autism. Mol Autism. 2014;5:40. https://doi.org/10.1186/2040-2392-5-40.
26.
Oh DH, Kim IB, Kim SH, Ahn DH. Predicting autism spectrum disorder using blood-based gene expression signatures and machine learning. Clin Psychopharmacol Neurosci. 2017;15:47–52. https://doi.org/10.9758/cpn.2017.15.1.47.
27.
Pramparo T, Pierce K, Lombardo MV, et al. Prediction of autism by translation and immune/inflammation coexpressed genes in toddlers from pediatric community practices. JAMA Psychiatry. 2015;72:386–94. https://doi.org/10.1001/jamapsychiatry.2014.3008.
28.
Anderson JS, Nielsen JA, Froehlich AL, et al. Functional connectivity magnetic resonance imaging classification of autism. Brain. 2011;134(Pt 12):3742–54. https://doi.org/10.1093/brain/awr263.
29.
Casanova MF. Neuropathological and genetic findings in autism: the significance of a putative minicolumnopathy. Neuroscientist. 2006;12:435–41. https://doi.org/10.1177/1073858406290375.
30.
Chen CP, Keown CL, Jahedi A, et al. Diagnostic classification of intrinsic functional connectivity highlights somatosensory, default mode, and visual regions in autism. Neuroimage Clin. 2015;8:238–45. https://doi.org/10.1016/j.nicl.2015.04.002.
31.
Ecker C, Rocha-Rego V, Johnston P, et al. Investigating the predictive value of whole-brain structural MR scans in autism: a pattern classification approach. Neuroimage. 2010;49:44–56. https://doi.org/10.1016/j.neuroimage.2009.08.024.
32.
Ghiassian S, Greiner R, Jin P, Brown MR. Using functional or structural magnetic resonance images and personal characteristic data to identify ADHD and autism. PLoS One. 2016;11:e0166934. https://doi.org/10.1371/journal.pone.0166934.
33.
Gori I, Giuliano A, Muratori F, et al. Gray matter alterations in young children with autism spectrum disorders: comparing morphometry at the voxel and regional level. J Neuroimaging. 2015;25:866–74. https://doi.org/10.1111/jon.12280.
34.
Ingalhalikar M, Kanterakis S, Gur R, Roberts TP, Verma R. DTI based diagnostic prediction of a disease via pattern classification. Med Image Comput Comput Assist Interv. 2010;13(Pt 1):558–65. https://doi.org/10.1007/978-3-642-15705-9_68.
35.
Ingalhalikar M, Parker D, Bloy L, Roberts TP, Verma R. Diffusion based abnormality markers of pathology: toward learned diagnostic prediction of ASD. Neuroimage. 2011;57:918–27. https://doi.org/10.1016/j.neuroimage.2011.05.023.
36.
Jiao Y, Chen R, Ke X, Chu K, Lu Z, Herskovits EH. Predictive models of autism spectrum disorder based on brain regional cortical thickness. Neuroimage. 2010;50:589–99. https://doi.org/10.1016/j.neuroimage.2009.12.047.
37.
Lange N, Dubray MB, Lee JE, et al. Atypical diffusion tensor hemispheric asymmetry in autism. Autism Res. 2010;3:350–8. https://doi.org/10.1002/aur.162.
38.
Lim L, Marquand A, Cubillo AA, et al. Disorder-specific predictive classification of adolescents with attention deficit hyperactivity disorder (ADHD) relative to autism using structural magnetic resonance imaging. PLoS One. 2013;8:e63660. https://doi.org/10.1371/journal.pone.0063660.
39.
Nielsen JA, Zielinski BA, Fletcher PT, et al. Multisite functional connectivity MRI classification of autism: ABIDE results. Front Hum Neurosci. 2013;7:599. https://doi.org/10.3389/fnhum.2013.00599.
40.
Plitt M, Barnes KA, Martin A. Functional connectivity classification of autism identifies highly predictive brain features but falls short of biomarker standards. Neuroimage Clin. 2014;7:359–66. https://doi.org/10.1016/j.nicl.2014.12.013.
41.
Shen MD, Nordahl CW, Young GS, et al. Early brain enlargement and elevated extra-axial fluid in infants who develop autism spectrum disorder. Brain. 2013;136(Pt 9):2825–35. https://doi.org/10.1093/brain/awt166.
42.
Uddin LQ, Supekar K, Lynch CJ, et al. Salience network-based classification and prediction of symptom severity in children with autism. JAMA Psychiatry. 2013;70:869–79. https://doi.org/10.1001/jamapsychiatry.2013.104.
43.
Varol E, Gaonkar B, Erus G, Schultz R, Davatzikos C. Feature ranking based nested support vector machine ensemble for medical image classification. Proc IEEE Int Symp Biomed Imaging. 2012:146–9. https://doi.org/10.1109/ISBI.2012.6235505.
44.
Wang L, Wee CY, Tang X, Yap PT, Shen D. Multi-task feature selection via supervised canonical graph matching for diagnosis of autism spectrum disorder. Brain Imaging Behav. 2016;10:33–40. https://doi.org/10.1007/s11682-015-9360-1.
45.
Wee CY, Wang L, Shi F, Yap PT, Shen D. Diagnosis of autism spectrum disorders using regional and interregional morphological features. Hum Brain Mapp. 2014;35:3414–30.
46.
Xiao X, Fang H, Wu J, et al. Diagnostic model generated by MRI-derived brain features in toddlers with autism spectrum disorder. Autism Res. 2017;10:620–30. https://doi.org/10.1002/aur.1711.
47.
Alabdali A, Al-Ayadhi L, El-Ansary A. Association of social and cognitive impairment and biomarkers in autism spectrum disorders. J Neuroinflammation. 2014;11:4. https://doi.org/10.1186/1742-2094-11-4.
48.
Al-Yafee YA, Al-Ayadhi LY, Haq SH, El-Ansary AK. Novel metabolic biomarkers related to sulfur-dependent detoxification pathways in autistic patients of Saudi Arabia. BMC Neurol. 2011;11:139. https://doi.org/10.1186/1471-2377-11-139.
49.
El-Ansary A, Al-Ayadhi L. Lipid mediators in plasma of autism spectrum disorders. Lipids Health Dis. 2012;11:160. https://doi.org/10.1186/1476-511X-11-160.
50.
El-Ansary AK, Bacha AG, Al-Ayahdi LY. Impaired plasma phospholipids and relative amounts of essential polyunsaturated fatty acids in autistic patients from Saudi Arabia. Lipids Health Dis. 2011;10:63. https://doi.org/10.1186/1476-511X-10-63.
51.
El-Ansary AK, Bacha AG, Al-Ayahdi LY. Plasma fatty acids as diagnostic markers in autistic patients from Saudi Arabia. Lipids Health Dis. 2011;10:62. https://doi.org/10.1186/1476-511X-10-62.
52.
Gevi F, Zolla L, Gabriele S, Persico AM. Urinary metabolomics of young Italian autistic children supports abnormal tryptophan and purine metabolism. Mol Autism. 2016;7:47. https://doi.org/10.1186/s13229-016-0109-5.
53.
Heyer NJ, Echeverria D, Woods JS. Disordered porphyrin metabolism: a potential biological marker for autism risk assessment. Autism Res. 2012;5:84–92. https://doi.org/10.1002/aur.236.
54.
Kuwabara H, Yamasue H, Koike S, et al. Altered metabolites in the plasma of autism spectrum disorder: a capillary electrophoresis time-of-flight mass spectroscopy study. PLoS One. 2013;8:e73814. https://doi.org/10.1371/journal.pone.0073814.
55.
Li SO, Wang JL, Bjørklund G, Zhao WN, Yin CH. Serum copper and zinc levels in individuals with autism spectrum disorders. Neuroreport. 2014;25:1216–20. https://doi.org/10.1097/WNR.0000000000000251.
56.
Momeni N, Bergquist J, Brudin L, et al. A novel blood-based biomarker for detection of autism spectrum disorders. Transl Psychiatry. 2012;2:e91. https://doi.org/10.1038/tp.2012.19.
57.
Nadal-Desbarats L, Aїdoud N, Emond P, et al. Combined 1H-NMR and 1H–13C HSQC-NMR to improve urinary screening in autism spectrum disorders. Analyst. 2014;139:3460–8. https://doi.org/10.1039/c4an00552j.
58.
Shimmura C, Suda S, Tsuchiya KJ, et al. Alteration of plasma glutamate and glutamine levels in children with high-functioning autism. PLoS One. 2011;6:e25340. https://doi.org/10.1371/journal.pone.0025340.
59.
Soria AC, Wright B, Goodall DM, Wilson J. Data processing in metabolic fingerprinting by CE-UV: application to urine samples from autistic children. Electrophoresis. 2007;28:950–64. https://doi.org/10.1002/elps.200600381.
60.
Testa C, Nuti F, Hayek J, et al. Di-(2-ethylhexyl) phthalate and autism spectrum disorders. ASN Neuro. 2012;4:223–9. https://doi.org/10.1042/AN20120015.
61.
Vargason T, Howsmon DP, Melnyk S, James SJ, Hahn J. Mathematical modeling of the methionine cycle and transsulfuration pathway in individuals with autism spectrum disorder. J Theor Biol. 2017;416:28–37. https://doi.org/10.1016/j.jtbi.2016.12.021.
62.
Wang H, Liang S, Wang M, et al. Potential serum biomarkers from a metabolomics study of autism. J Psychiatry Neurosci. 2016;41:27–37. https://doi.org/10.1503/jpn.140009.
63.
West PR, Amaral DG, Bais P, et al. Metabolomics as a tool for discovery of biomarkers of autism spectrum disorder in the blood plasma of children. PLoS One. 2014;9:e112445. https://doi.org/10.1371/journal.pone.0112445.
64.
Xiong X, Liu D, Wang Y, Zeng T, Peng Y. Urinary 3-(3-hydroxyphenyl)-3-hydroxypropionic acid, 3-hydroxyphenylacetic acid, and 3-hydroxyhippuric acid are elevated in children with autism spectrum disorders. Biomed Res Int. 2016;2016 9485412. https://doi.org/10.1155/2016/9485412.
65.
Yan CL, Zhang J, Hou Y. Decreased plasma levels of lipoxin A4 in children with autism spectrum disorders. Neuroreport. 2015;26:341–5. https://doi.org/10.1097/WNR.0000000000000350.
66.
Al-Ayadhi LY, Ben Bacha AG, Kotb M, El-Ansary AK. A novel study on amyloid beta peptide 40, 42 and 40/42 ratio in Saudi autistics. Behav Brain Funct. 2012;8:4. https://doi.org/10.1186/1744-9081-8-4.
67.
Cai J, Ding L, Zhang JS, Xue J, Wang LZ. Elevated plasma levels of glutamate in children with autism spectrum disorders. Neuroreport. 2016;27:272–6. https://doi.org/10.1097/WNR.0000000000000532.
68.
Chen J, Xin K, Wei J, Zhang K, Xiao H. Lower maternal serum 25(OH) D in first trimester associated with higher autism risk in Chinese offspring. J Psychosom Res. 2016;89:98–101. https://doi.org/10.1016/j.jpsychores.2016.08.013.
69.
El-Ansary AK, Ben Bacha AG, Al-Ayadhi LY. Proinflammatory and proapoptotic markers in relation to mono and di-cations in plasma of autistic patients from Saudi Arabia. J Neuroinflammation. 2011;8:142. https://doi.org/10.1186/1742-2094-8-142.
70.
Pagan C, Delorme R, Callebert J, et al. The serotonin-N-acetylserotonin-melatonin pathway as a biomarker for autism spectrum disorders. Transl Psychiatry. 2014;4:e479. https://doi.org/10.1038/tp.2014.120.
71.
Wang J, Zou Q, Han R, Li Y, Wang Y. Serum levels of glial fibrillary acidic protein in Chinese children with autism spectrum disorders. Int J Dev Neurosci. 2017;57:41–5. https://doi.org/10.1016/j.ijdevneu.2017.01.004.
72.
Wang M, Chen H, Yu T, Cui G, Jiao A, Liang H. Increased serum levels of brain-derived neurotrophic factor in autism spectrum disorder. Neuroreport. 2015;26:638–41. https://doi.org/10.1097/WNR.0000000000000404.
73.
Zhang QB, Gao SJ, Zhao HX. Thioredoxin: a novel, independent diagnosis marker in children with autism. Int J Dev Neurosci. 2015;40:92–6. https://doi.org/10.1016/j.ijdevneu.2014.11.007.
74.
Zhang QB, Jiang LF, Kong LY, Lu YJ. Serum brain-derived neurotrophic factor levels in Chinese children with autism spectrum disorders: a pilot study. Int J Dev Neurosci. 2014;37:65–8. https://doi.org/10.1016/j.ijdevneu.2014.06.013.
75.
Zhao HX, Yin SS, Fan JG. High plasma neopterin levels in Chinese children with autism spectrum disorders. Int J Dev Neurosci. 2015;41:92–7. https://doi.org/10.1016/j.ijdevneu.2015.02.002.
76.
Chan AS, Sze SL, Cheung MC. Quantitative electroencephalographic profiles for children with autistic spectrum disorder. Neuropsychology. 2007;21:74–81. https://doi.org/10.1037/0894-4105.21.1.74.
77.
Duffy FH, Als H. A stable pattern of EEG spectral coherence distinguishes children with autism from neuro-typical controls - a large case control study. BMC Med. 2012;10:64. https://doi.org/10.1186/1741-7015-10-64.
78.
Eldridge J, Lane AE, Belkin M, Dennis S. Robust features for the automatic identification of autism spectrum disorder in children. J Neurodev Disord. 2014;6:12. https://doi.org/10.1186/1866-1955-6-12.
79.
Jamal W, Das S, Oprescu IA, Maharatna K, Apicella F, Sicca F. Classification of autism spectrum disorder using supervised learning of brain connectivity measures extracted from synchrostates. J Neural Eng. 2014;11:046019. https://doi.org/10.1088/1741-2560/11/4/046019.
80.
Karhson DS, Golob EJ. Atypical sensory reactivity influences auditory attentional control in adults with autism spectrum disorders. Autism Res. 2016;9:1079–92. https://doi.org/10.1002/aur.1593.
81.
Matlis S, Boric K, Chu CJ, Kramer MA. Robust disruptions in electroencephalogram cortical oscillations and large-scale functional networks in autism. BMC Neurol. 2015;15:97. https://doi.org/10.1186/s12883-015-0355-8.