Profiles of urine and blood metabolomics in autism spectrum disorders

Abstract

Early diagnosis and treatment for autism spectrum disorder (ASD) pose challenges. The current diagnostic approach for ASD is mainly clinical assessment of patient behaviors. Biomarkers-based identification of ASD would be useful for pediatricians. Currently, there is no specific treatment for ASD, and evidence for the efficacy of alternative treatments remains inconclusive. The prevalence of ASD is increasing, and it is becoming more urgent to find the pathogenesis of such disorder. Metabolomic studies have been used to deeply investigate the alteration of metabolic pathways, including those associated with ASD. Metabolomics is a promising tool for identifying potential biomarkers and possible pathogenesis of ASD. This review comprehensively summarizes and discusses the abnormal metabolic pathways in ASD children, as indicated by evidence from metabolomic studies in urine and blood. In addition, the targeted interventions that could correct the metabolomic profiles relating to the improvement of autistic behaviors in affected animals and humans have been included. The results revealed that the possible underlying pathophysiology of ASD were alterations of amino acids, reactive oxidative stress, neurotransmitters, and microbiota-gut-brain axis. The potential common pathways shared by animal and human studies related to the improvement of ASD symptoms after pharmacological interventions were mammalian-microbial co-metabolite, purine metabolism, and fatty acid oxidation. The content of this review may contribute to novel biomarkers for the early diagnosis of ASD and possible therapeutic paradigms.

Introduction

There are concerns about the increasing prevalence of ASD in recent decades (Hansen et al. 2015). The underlying pathogenesis of ASD remains unclear, and there is no definitive treatment for ASD (Emberti Gialloreti and Curatolo 2018). Several hypotheses have proposed that many factors might cause ASD, and the most well-known are genetic factors. There might also be interactions between genetic and environmental factors or epigenetics (Chaste and Leboyer 2012). In addition, some prenatal factors have been shown to increase the risk of ASD, including prenatal infection and exposure to specific drugs or toxins (Dietert et al. 2011). Interestingly, advanced technology-based genetic testing has revealed an abundance of candidate genes relating to ASD (da Silva Montenegro et al. 2020). Although ASD phenotypes can be seen in some single-gene disorders and metabolic diseases, only a few identifiable causes of ASD can be detected by genetic or metabolic testing. Therefore, the majority of cases are idiopathic ASD. With the lack of gold-standard laboratory tests for ASD, early diagnosis and initial intervention of such disorder are challenging (Huerta and Lord 2012).

Due to the complexity of ASD, the target approach for diagnosis is shifting to omics studies, including metabolomics. Due to its ability to measure metabolic profiles influenced by genetic and environmental factors, a metabolomic study may reveal the cause of abnormal metabolic pathways and identify possible biomarkers for diagnosis in various diseases, including ASD (Glinton and Elsea 2019).

An accumulation of evidence has indicated that the metabolomic profiles in biological samples were different between autistic and typically developing children (Anwar et al. 2018; Bitar et al. 2018; Chen et al. 2019; Cozzolino et al. 2014; Delaye et al. 2018; Diémé et al. 2015; Emond et al. 2013; Gevi et al. 2016; Kelly et al. 2019; Kuwabara et al. 2013; Liu et al. 2019; Lussu et al. 2017; Lv et al. 2018; Mavel et al. 2013; Ming et al. 2012; Nadal-Desbarats et al. 2014; Noto et al. 2014; Orozco et al. 2019; Rangel-Huerta et al. 2019; Smith et al. 2019; Wang et al. 2016; West et al. 2014; Xiong et al. 2019, 2016; Yap et al. 2010). The affected pathways may be responsible for the mechanisms underlying ASD. Although there are several metabolomic studies in ASD, a few literatures have focused directly on perturbed metabolic pathways. This review will comprehensively discuss and summarize the abnormal metabolic pathways in ASD patients from the previous metabolomic studies. Alterations of metabolomic profiles after receiving the pharmacological interventions from animal and human studies are also reviewed.

Abnormal metabolism associated with ASD from metabolomic studies in urine samples

The abnormal metabolisms associated with ASD from metabolomic studies in urine samples are listed in Table 1. These abnormalities reflected disturbances of various metabolic pathways. In the case of the pathways of oxidative damage, the oxidative stress markers were found to be positively correlated with the degrees of ASD severity (Damodaran and Arumugam 2011). Previous studies demonstrated that lower levels of antioxidants such as carnosine and anserine occurred in children with ASD compared to those of typical children (Liu et al. 2019; Ming et al. 2012; Yui et al. 2017). An increase in the levels of oxidative stress via nitric oxide production could be indicated by an excessive amount of arginine (Delwing et al. 2008), as demonstrated by the higher levels of N-acetylarginine in ASD compared to that of typical children (Diémé et al. 2015; Olesova et al. 2020). The proposed mechanism of oxidative stress might be implicated in the pathophysiology of ASD.

Table 1.

Abnormal metabolism associated with autism spectrum disorder from metabolomic studies in urine samples

Ref and Subjects	Age (years)	Method	Categories of Metabo- lite/ Metabolic pathway involvement	Changes in metabolites compared to typical children		Interpretation
				Increase	Decrease
Gevi et al. 2020
40 ASD compared with 40 typical children	3-8	Untargeted approach using UHPLC-MS	Neurotransmitter	Glutamate	GABA	There are abnormalities of neurotransmitters and metabolites related to microbiota-brain-gut axis in the urine of children with ASD.
					Pyridoxal phosphate
			Neurotransmitter/microbiota-brain-gut axis	4-Cresol	Noradrenaline
				Ascorbate	Adrenaline
				Homovanillic acid	MHPG Vanillylmandelic acid
Liang, Ke et al. 2020
40 ASD compared with 40 typical children	3-12	Untargeted approach using UHPLC-QTOF/MS	Nicotinate and nicotinamide metabolism	Nicotinamide		There are abnormalities of nicotinamide, phosphorylcholine, amino acid, acetyl-CoA synthase, lysine, nucleosides, neurotransmitter, naphthalene metabolites, sialic acid, purine, and lipid metabolism in urine of children with ASD.
			Phosphorylcholine metabolism	Phosphorylcholine
			Amino acid metabolism	Glycylglutamate Alanylthreonine Threonylaspartate Histidylproline Lysylproline Valylmethionine Methionylglutamine	Prolylserine
			Acetyl-CoA synthase	Acetylcarnitine
			Tryptophan kynurenine pathway	D-neopterin 7,8-Dihydroneopterin Anthranilic acid
			Catabolism of lysine	5-Aminopentanoic acid
			Modified nucleosides	1-Methyladenosine 3’-O-Methylinosine
			Neurotransmitter	N-Acetylaspartylglutamate
			Naphthalene metabolites	1-Naphthol
			Sialic acid pathway	N-Acetylneuraminic acid
			Purine metabolism	Deoxyinosine
			Lipid metabolism	Behenic acid
			Other	Bethanechol cation S-Methyl-5’-thioadenosine
Liang, Xiao et al. 2020
22 ASD compared with 22 non-ASD siblings	3-9	Untargeted approach using 1H-NMR	Cysteine metabolism	Taurine		There are abnormalities of cysteine, methionine, oxidative stress, sulfur, and tryptophan-serotonin-melatonin metabolic pathway in urine of children with ASD.
			Methionine cycle	Taurine
			Reactive oxidative stress	Taurine
			Sulfur metabolism	Taurine
			Tryptophan-serotonin-melatonin pathway	Tryptophan Serotonin	Melatonin
			Others	Creatine	Butyrate Citrate Lactate Pantothenate Trigonelline
Mussap et al. 2020
31 ASD compared with 26 typical children	2-11	Untargeted approach using GC-MS	Oxidative stress	Cystine	7-Methylxanthine Uric acid	There are differences of metabolite levels resulting from oxidative stress, mitochondrial dysfunction, sugar metabolism, gut dysbiosis, and diet in urine of children with ASD compared to those of typically developing children.
			Mitochondrial dysfunction	Lactic acid
			Sugar metabolism		Scylloinositol
			Gut dysbiosis	Quinic acid Hippuric acid Tryptophan Indole-3-acetic acid Allyl thioacetic acid Leucine
			Diet	1-Methylhistidine
			Others		Aminomalonic acid
Olesova et al. 2020
24 ASD compared with 13 typical children	6-10	Targeted approach using UHPLC-QqQ-MS/MS	Reactive oxidative stress	Methylguanidine N-acetylarginine		There are abnormalities of reactive oxidative stress and gut bacteria metabolism in the urine of children with ASD.
			Gut bacteria metabolism	Indoxyl sulphate Indole-3-acetic acid
Liu et al. 2019
57 ASD compared with 81 typical children	2-12	Targeted approach using LC-MS/MS	Ornithine (Urea) cycle Ornithine (Urea) cycle	Ornithine-to-Citrulline ratio	Proline	There are abnormalities of ornithine (urea) cycle, methionine, lysine, reactive oxidative stress, tryptophan-serotonin metabolism in urine of children with ASD.
				Arginine-to-Ornithine	Citrulline
				Arginine 4-Hydroxyproline	Aspartic acid
			Methionine cycle	Methionine sulfoxide	Homocysteine 5-Aminovaleric acid a-Aminoadipic acid Ethanolamine
			Lysine metabolism		Lysine
			Reactive oxidative stress		Anserine Carnosine
			Tryptophan-Serotonin metabolism	5-Hydroxytryptamine
			Others	2-Aminoisobutyric acid
Chen et al. 2019
156 ASD compared with 64 typical children	NA	Untargeted approach using GC/MS	Phenylalanine metabolism	Phenylactic acid		There are abnormalities of phenylalanine, TCA cycle, bone metabolism, mammalian-microbial co-metabolism in urine of ASD.
			TCA cycle		Aconitic acid Carboxycitric acid
			Bone metabolism		Phosphoric acid
			Mammalian-microbial co-metabolism		3-Oxoglutaric acid Carboxycitric acid
			Others	3-Hydroxy-3-methylglutaric	Fumaric
				Creatinine	N-Acetylcysteine
				Oxalic	Malonic
				Pyruvic	Tricarballylic
				4-cresol	Glycolic
				2-hydroxybutyric	Malic Tartaric 3-Hydroxyglutaric 2-Oxoglutaric
Xiong et al. 2019
51 ASD compared with 51 typical children	3-7	Untargeted approach using HPLC-QTOF-MS	Adenosine-Adenine pathway	Adenine		There are abnormalities of adenosine-adenine pathway, methylation regulation of retinoic acid-RORA pathway, and arginine metabolism, predominantly in female patient with ASD.
			Methylation regulation of retinoic acid-RORA pathway	2-Methylguanosine 7alpha-Hydroxytestololactone
			Arginine metabolism	Creatinine	Creatine
Bitar et al. 2018
40 ASD compared with 40 typical children	NA	Untargeted approach using 1H-NMR and LC-MS	Glycine, serine and threonine metabolism	Phosphoserine	Threonine Creatine Serine	There are abnormalities in amino acids, nicotinic acid, TCA cycle, purine, vitamin, and fatty acid metabolism in urine of ASD.
			Phenylalanine metabolism		N-acetylphenylalanine Tyrosine Hydroxybenzoic acid
			Glutamate, Arginine and Proline metabolism	Glutamic acid	Creatine Hydroxyproline
			Histidine metabolism	Glutamic acid	Urocanic acid
			Cysteine and Methionine metabolism	Phosphoserine	Cysteic acid Serine
			Propanoate metabolism		2-hydroxybutyric acid
			Nicotinate and nicotinamide metabolism	Nicotinamide ribotide Trigonelline
			TCA cycle		Citric acid
			Purine metabolism	5-amino-imidazole-4-carboxamide	Guanine
			Vitamin B6 metabolism	Riboflavin
			Fatty acid oxidation		N-amidino aspartic acid Acetylcarnitine
			Others	Glycerol-3-phosphate Cholic acid	Methyl acetoacetic acid
Lussu et al. 2017
21 ASD compared with 21 non-ASD siblings	4-16	Untargeted approach using 1H-NMR	Neurotransmitter	Glycine	Glutamate	There are abnormalities of neurotransmitter, tryptophan-serotonin, mammalian-microbial co-metabolism, oxidative stress metabolism in urine of ASD.
			Tryptophan-serotonin metabolism	Tryptophan
			Mammalian-microbial co-metabolism	Hippurate
			Oxidative stress		Taurine Lactate
			Others	Creatine D-threitol	Valine Betaine Creatinine
Xiong et al. 2016
62 ASD compared with 62 typical children	1.5-7	Untargeted approach using GC/MS	Mammalian-microbial co-metabolism	3-(3-hydroxyphenyl)-3-Hydroxypropionic acid 3-Hydroxyphenylacetic acid 3-Hydroxyhippuric acid		There are abnormalities in mammalian-microbial co-metabolism in urine of ASD.
Gevi et al. 2016
30 ASD compared with 30 typical children	2-7	Untargeted approach using LC-MS	Purine metabolism	Inosine Hypoxanthine Xanthosine	Adenosine diphosphate	There are abnormalities of nucleic acid (purine, pyrimidine), amino acid (tryptophan), vitamins (B1, B2, B6, and pantothenate), mammalian-microbial co-metabolism, glutathione, pentose phosphate pathway, benzoate degradation, and carbohydrate metabolism in urine of ASD.
			Pyrimidine metabolism	Uridine
			Tryptophan metabolism	Xanthurenic acid Quinolinic acid Tryptophan	Kynurenine
			Mammalian-microbial co-metabolism	Indolyl 3-acetic acid Indolyl lactate
			Disaccharide metabolism	Trehalose/sucrose	Cellobiose
			Other amino acid metabolism	Phenylalanine Histidine	Methionine
			Glutathione metabolism		Pyroglutamic acid
			Pentose Phosphate Pathway	6-phospho-D-gluconic acid Ribose
			Benzoate degradation	p-cresol	p-hydroxybenzoate
			Vit B1 metabolism	Thiamine
			Vit B2 metabolism	Riboflavin
			Vit B6 metabolism	4-pyriodic acid
			Pantothenate and CoA metabolism		Glucose-6-phosphate
			Glycolysis		Glucose-6-phosphate
Diémé et al. 2015
30 ASD compared with 32 typical children	NA	Untargeted approach using 1H-NMR, 1H-13C-NMR and LC-HRMS	Reactive oxidative stress	N-acetylarginine	Methylguanidine Guanidinosuccinic acid	There are abnormalities in reactive oxidative stress, tyrosine, pyrimidine, and mammalian-microbial co-metabolism in urine of ASD.
			Tyrosine metabolism	Dihydroxy-1H-indole glucuronide I
			Pyrimidine metabolism		Dihydrouracil
			Mammalian-microbial co-metabolism	Indoxyl Indoxyl sulfate Alpha-N-Phenylacetyl-L-glutamine p-cresol sulfate	Desaminotyrosine
			Others	N-Acetylasparagine Valine Glucuronic acid
Cozzolino et al. 2014
24 ASD compared with 21 typical children	6.9 +/−2.1	Untargeted approach using Solid-phase microextraction with GC/MS	Leucine metabolism	3-Methylbutanal (under acid conditions)		There are abnormalities in leucine, fatty acid, and degradation of carbohydrate metabolism in urine of ASD.
			Peroxidation of fatty acid	3-Methylbutanal (under acid conditions) 2-Methylbutanal (under acid conditions)
			Dietary degradation of carbohydrate		2-Methylmercaptofuran (under acid conditions) 2-Pentylfuran (under acid conditions)
			Others	3-Methylcyclopentanone (under acid conditions)	2-Heptanone (under alkaline conditions)
				Hexane (under acid conditions)	Ethanol (under acid conditions)
				2,3-Dimethylpyrazine (under alkaline conditions)	Dimethyl trisulfide (under acid conditions)
				2-Methylpyrazine (under alkaline conditions)	Methoxy-phenyloxime (under alkaline conditions)
				Isoxazole (under alkaline conditions)	3-Ethylpyridine (under alkaline conditions) Acetophenone (under alkaline conditions)
Noto et al. 2014
21 ASD compared with 21 non-ASD siblings	4-16	Untargeted approach using GC/MS	Mammalian-microbial co-metabolism	3-(3-Hydroxyphenyl)-3-hydroxypropanoic acid Glycolic acid 4-Hydroxyhippuric acid		There are abnormalities of TCA cycle, Tryptophan, Glutathione, oxidative cleavage of N-acetylglucosamine, Tyrosine, pentose phosphate pathway, and mammalian-microbial co-metabolism in urine of ASD.
			TCA cycle	Aconitic acid
			Tryptophan metabolism	Tryptophan
			Glutathione metabolism	Pyroglutamic acid
			Oxidative cleavage of N-acetylglucosamine	Erythronic acid
			Tyrosine pathway	Phenylalanine Tyrosine 4-Hydroxyphenylacetic acid Homovanillic acid
			Pentose Phosphate Pathway	Ribose Arabinofuranose Threitol Polyols arabitol Xylitol
			From dietary sources		Fructose 1,2,3-Butanetriol Propylene glycol
Mavel et al. 2013
30 ASD compared with 28 typical children	6-14	Targeted approach using 1H-NMR	Organic acid	Succinic acid		There are abnormalities of organic acid and amino acid metabolism in urine of ASD.
			Amino acid	Glycine	Creatine
				β-alanine Taurine	3-methylhistidine
Emond et al. 2013
26 ASD compared with 24 typical children	6-14	Untargeted approach using GC/MS	Mammalian-microbial co-metabolism		1H-indole-3-acetate Hippurate 3-hydroxyhippurate	There are abnormalities of organic acid, fatty acid, phenols, and mammalian-microbial co-metabolism in urine of ASD.
			Organic acid	Succinate Glycolate
			Fatty acid		Palmitate Stearate 3-methyladipate
			Phenols		p-Hydroxy mandelate 3-hydroxyphenylacetate
Ming et al. 2012
48 ASD compared with 53 typical children	6-14	Untargeted approach combined using UPLC-MS/MS and GC/MS	Histidine catabolism	Trans-Urocanate		There are abnormalities of amino acid (Histidine, Lysine, Tryptophan, Leucine), reactive oxidative stress, and mammalian-microbial co-metabolism in urine of ASD.
			Lysine and tryptophan catabolism	Glutaroylcarnitine
			Leucine catabolism	3-Methylglutaroylcarnitine
			Mammalian-microbial co-metabolism	2-(4-Hydroxyphenyl) propionate	3-(3-Hydroxyphenyl) propionate
				Taurocholenate sulfate	5-Amino-valerate
			Reactive oxidative stress		Carnosine Urate
			Amino acid and gamma-glutaryl amino acids		Glycine Serine Threonine Alanine β-alanine Histidine Taurine N-acetylglycine Gamma-glutamylleucine Gamma-glutamyltyrosine Gamma-glutamylthreonine
Yap et al. 2010
39 ASD compared with 28 non-ASD siblings 34 typical children	3-9	Untargeted approach using 1H-NMR	Nicotinic acid metabolism	N-methyl-2-pyridone-5-carboxamide N-methyl nicotinic acid N-methyl nicotinamide		There are abnormalities of nicotinic acid, amino acid, mammalian microbial co-metabolism in urine of ASD
			Amino acid metabolism	Taurine N-acetyl glycoprotein fragment Succinate	Glutamate
			Mammalian microbial co-metabolites	Dimethylamine	Hippurate Phenylacetylglutamine
Nadal-Desbarats et al. 2014
30 ASD compared with 28 typical children	6-14	Targeted approach using combined 1H-NMR and 2D 1H-13C HSQC NMR	TCA cycle	Succinate		There are abnormalities of TCA cycle and glutamate metabolism in urine of ASD.
			Glutamate metabolism		Glutamate
			Others		3-methylhistidine

ASD autism spectrum disorder, TCA tricarboxylic acid cycle, UPLC-MS/MS ultra-performance liquid chromatography—tandem mass spectrometer, UHPLC-MS ultra-high-performance liquid chromatography-mass spectrometry, UHPLC-QTOF/MS ultra-high-performance liquid chromatography-quadrupole time-of-flight mass spectrometry, UHPLC-QqQ-MS/MS ultra-high-performance liquid chromatography-tandem triple quadrupole-mass spectrometry, LC–MS liquid chromatography–mass spectrometry, LC–MS/MS liquid chromatography–tandem mass spectrometry, LC-HRMS liquid chromatography–high resolution mass spectrometry, GC–MS gas chromatography–mass spectrometry, ¹H-NMR hydrogen-1 nuclear magnetic resonance, ¹H-¹³C-NMR hydrogen-1 and carbon-13 nuclear magnetic resonance, 2D ¹H-¹³C HSQC NMR two-dimensional hydrogen-1 and carbon-13 heteronuclear single quantum correlation nuclear magnetic resonance; MS/MS: tandem mass spectrometry, HPLC-QTOF-MS high-performance liquid chromatography-quadrupole-time of flight mass spectrometry, GABA gamma-aminobutyric acid, RORA related orphan receptor alpha, NA data not available

Mammalian-microbial co-metabolism was also found to be associated with ASD symptoms. The comparison of the urine metabolite levels between children with autism and typical development showed the reduced levels of carboxycitric acid and 3-oxoglutaric acid (Chen et al. 2019), while the levels of hippurate (Kałużna-Czaplińska 2011; Lussu et al. 2017), 3-(3-hydroxyphenyl)-3-hydroxypropionic acid (HPHPA) (Noto et al. 2014; Xiong et al. 2016), 3-hydroxyphenylacetic acid (3HPA) (Xiong et al. 2016), 3-hydroxyhippuric acid (3HHA) (Xiong et al. 2016), indole-3-acetic acid (Gevi et al. 2016; Olesova et al. 2020), indolyl lactate (Gevi et al. 2016), and glycolic acid (Noto et al. 2014) were increased in children with ASD compared to children with typical development. However, these results were inconsistent with some studies (Emond et al. 2013; Yap et al. 2010), in which the decreased levels of hippurate was shown. These contrary results could be as a result of unrestricted food among participants prior to the sample collection and heterogeneity of ASD.

Abnormalities of various pathways associated with amino acids, including tryptophan, arginine, phenylalanine, tyrosine, leucine, histidine, and methionine were also observed in cases of ASD. Tryptophan has been the most discussed amino acid associated with the development of ASD. Liang et al. (2020b) showed evidence supporting the association between tryptophan-serotonin-melatonin pathways and autism. In fact, they found increased tryptophan and serotonin levels while decreased melatonin in the urine of children with ASD (Liang et al. 2020b). This finding was in line with the dysfunction of melatonin that leads to sleep–wake rhythm disturbance in patients with ASD (Wu et al. 2020). Another study by Liang et al. (2020a, b) was found an increase in 7,8-dihydroneopterin and neopterin in the urine of children with ASD. The increase in these two metabolites, which are the markers of reactive oxygen species, were proposed to be related to the disruption of tryptophan kynurenine pathway and pathogenesis of autism (Liang et al. 2020a, b). Prior studies observed a higher concentration of 5-hydroxytryptamine (Liu et al. 2019), a downstream metabolite of tryptophan, and tryptophan (Gevi et al. 2016; Lussu et al. 2017; Noto et al. 2014) in the urine samples of children with ASD compared to that of typical children. Additionally, the disturbance of tryptophan metabolism was revealed by the alteration of multiple tryptophan intermediates, including xanthurenic acid, quinolinic acid, and kynurenine in the urine samples of ASD cases (Gevi et al. 2016). Abnormalities in arginine metabolism were also exhibited, as indicated by the increased levels of creatinine while decreased levels of creatine in the urine of female children with ASD (Xiong et al. 2019). The abnormalities regarding arginine pathway-related enzymes and receptors could be explained by creatine is one of the byproducts of this pathway. Abnormalities in phenylalanine metabolism were shown as the higher excretion of phenylactic acid and phenylalanine in the urine of children with ASD compared to that of typical children (Chen et al. 2019; Gevi et al. 2016; Noto et al. 2014). Disturbance of tyrosine metabolism was also proposed in some studies, as alterations of tyrosine intermediates in children with ASD were found (Diémé et al. 2015; Noto et al. 2014). Leucine metabolism was also assumed to be dysfunctional, as indicated by the increased levels of 3-methylbutanal, a leucine intermediate, in children with ASD (Cozzolino et al. 2014). Histidine catabolic dysfunction in cases of ASD was also revealed in a study by Ming et al. (2012). Indeed, they found the increased levels of a histidine intermediate–urocanate in children with ASD compared to the controls (Ming et al. 2012). This was supported by the discovery of urocanase deficiency, which leads to excessive excretion of urocanic acid and manifestations of developmental delay (Espinós et al. 2009). Abnormalities in the methionine cycle were found as the levels of methionine sulfoxide increased and homocysteine decreased in the urine of children with ASD compared to those of typically developing children (Liu et al. 2019). Those two metabolites act as methyl group donors; therefore, these findings supported the association between abnormal methylation and the development of ASD (Behnia et al. 2015; Hannon et al. 2018; Ladd-Acosta et al. 2014).

There are evidence links between abnormalities of amino acid metabolism and the imbalance of neurotransmitters. Decreased lysine and ethanolamine levels were exhibited in urine from children with ASD and were thought to be associated with the lower levels of the excitatory neurotransmitters glutamate and acetylcholine (Liu et al. 2019). On the other hand, increased levels of 2-aminoisobutyric acid which is assumed to play a similar role as an inhibitory neurotransmitter, gamma aminobutyric acid (GABA), were observed in the urine of children with ASD compared to typically developing children (Liu et al. 2019). Decrease in excitatory neurotransmitter, glutamate, but increases in the inhibitory neurotransmitter, glycine, have also been demonstrated in several studies (Lussu et al. 2017; Nadal-Desbarats et al. 2014; Xiong et al. 2016). In contrast with other prior studies, an increase in urine glutamate were discovered in children with ASD (Bitar et al. 2018; Gevi et al. 2020). A reduction of urine GABA in children with ASD was explained by the possible link of the microbiota-gut-brain axis. The increased 4-cresol, a consequence of gut dysbiosis, inhibits dopamine β-hydroxylase that is responsible for converting dopamine to noradrenaline. Therefore, the accumulation of dopamine results in the reduction of GABA (Gevi et al. 2020). These inconsistent results could be due to the different ethnicities and spectrometric techniques among the studies.

Regarding the urea cycle pathway, there was a positive association between ASD and blood ammonia (Cohen 2002; Wang et al. 2012). The proposed mechanism responsible for the elevation of ammonia was an increase in the conversion of proline to 4-hydroxyproline by the enzyme pyrroline-5-carboxylate reductase (Mitsubuchi et al. 2008). In addition, the dysfunction of the enzymes arginine transcarbamylase and ornithine transcarbamylase led to the malfunction of the urea cycle pathway, as Liu et al. (2019) showed by demonstrating an increased metabolic ratio of ornithine-to-citrulline and arginine-to-ornithine in the urine of children with ASD.

It has been speculated that abnormalities in the TCA cycle was one of the potential mechanisms to be related to ASD. Previous studies had revealed a decrease in aconitic acid, carboxycitric acid, and citric acid in the urine samples of children with ASD compared with those of typical children (Bitar et al. 2018; Chen et al. 2019). On the other hand, another study reported increased cis-aconitic acid levels in children with ASD (Noto et al. 2014). Based on these findings, decreased levels of such metabolites were believed to be associated with aconitase deficiency (Noto et al. 2014). The inconsistent findings of aconitic levels were possibly due to the difference in ethnicity.

Abnormalities in purine metabolism, such as the adenosine-adenine pathway, in cases of ASD have been reported. These abnormalities were evidenced by increased levels of aminoimidazole (Bitar et al. 2018), inosine, hypoxanthine, and xanthosine (Gevi et al. 2016), but decreased levels of guanine (Bitar et al. 2018) and dihydrouracil (Diémé et al. 2015) in children with ASD compared to typically developing children.

There is some evidence to show alteration of both nicotinate and nicotinamide metabolism in cases of ASD. Indeed, previous studies demonstrated increased levels of nicotinamide (Bitar et al. 2018; Liang et al. 2020a, b; Yap et al. 2010) in urine compared to that of typically developing children. Excessive excretion of such metabolites was thought to cause damage to neurons and had been previously implicated in Parkinson’s disease (Willets et al. 1993; Williams et al. 1993). An increased nicotinic acid level has also been related to ASD (Melke et al. 2008). This alteration could be due to diminished conversion of tryptophan to serotonin by reduced acetylserotonin methyltransferase enzyme activity (Melke et al. 2008).

With regard to the abnormalities of carbohydrate metabolism, a study proposed that the lower activity of intestinal carbohydrate degradation enzymes might lead to the lower levels of furans in cases of ASD (Cozzolino et al. 2014). There was the first-time reporting of a significant reduction of urine scylloinositol in children with ASD (Mussap et al. 2020). A similar metabolome pattern of scylloinositol was also found in the brain of rats with ASD-like features (Toczylowska et al. 2020). These two studies might support the evidence of the association between inositol metabolism and autism. There are also abnormalities in the pentose phosphate pathway, as indicated by increased levels of pentose phosphate pathway-related metabolites such as ribose and arabinofuranose in children with ASD (Noto et al. 2014).

The alteration of lipid peroxidation could explain increased aldehydes such as 3-methylbutanal and 2-methylbutanal in ASD patients, because such aldehydes can derive from free radical-mediated lipid peroxidation (Cozzolino et al. 2014). A decrease in acetylcarnitine in urine samples of children with ASD (Bitar et al. 2018) supports the alteration of fatty acid oxidation. Also, the decreased level of acetylcarnitine could lead to impaired energy transportation. Thus, some ASD behaviors can be improved by L-carnitine supplementation (Frye 2015).

The distinguishable set of metabolites by metabolomics study in urine mentioned above could explain the underlying abnormal metabolic pathways in patients with ASD including the antioxidants or oxidative stress pathway, mammalian-microbial co-metabolism pathway, metabolism of some amino acids such as tryptophan and phenylalanine. Interestingly, the metabolite of such metabolic pathways might be possible biomarkers for ASD diagnosis that needed to be further investigated.

Abnormal metabolism associated with ASD from metabolomic studies in blood samples

The abnormal metabolic pathways associated with ASD from metabolomic studies in blood samples are listed in Table 2. The abnormalities pertinent to various amino acids, including tryptophan, tyrosine, and branched-chain amino acid pathways in blood samples have been reported as being associated with ASD. Abnormal tryptophan metabolism was indicated by decreased levels of N-formylanthranilic acid, a downstream metabolite of the tryptophan-derived kynurenine pathway, but increased serotonin levels in children with poor communication skills compared to those with typical skills (Kelly et al. 2019). Perturbation in tyrosine metabolism was also found to be correlated with ASD. A previous study revealed that blood levels of N-formylphenylalanine, which is a tyrosine intermediate, of children with a low communication score were lower than that of children with a typical score (Kelly et al. 2019). Furthermore, alterations in the metabolism of branched-chain amino acids (BCAA) were found in patients with ASD, as indicated by higher blood levels of alpha-keto derivatives from BCAA catabolism (Rangel-Huerta et al. 2019). The increase in ratios of specific amino acids such as glutamine, glycine, and ornithine with BCAA was reported in children with ASD (Smith et al. 2019). These abnormalities were assumed to be caused by a deficiency in the branched-chain ketoacid dehydrogenase kinase enzyme (Smith et al. 2019). The dysfunction of this enzyme eventually leads to a lower concentration of BCAA in plasma, resulting in neurological impairment (Novarino et al. 2012).

Table 2.

Abnormal metabolism associated with autism spectrum disorder from metabolomic studies in blood samples

Ref and Subjects	Age (years)	Method	Categories of Metabo- lites/ Involved metabolic pathways	Changes in metabolites compared to typical children		Interpretation
				Increase	Decrease
Kelly et al. 2019
403 children (365 categorized to “on schedule” and 38 categorized to “requiring further monitoring/evaluation” from ASQ)	3	Targeted approach using UPLC-MS/MS	Tyrosine metabolism		N-formylphenylalanine	There are abnormalities of amino acid (tyrosine, tryptophan, arginine, proline, methionine, cysteine, and taurine), lipid (phospholipid, sphingolipid, and fatty acid), urea cycle, and xenobiotics metabolism in children with poor ASQ communication skills, relative to children with typical communication development.
			Phospholipid metabolism	trimethylamine N-oxide
			Food Component/Plant	Cinnamoylglycine Erythritol	Pyrraline
			Endocannabinoid	Oleoyl ethanolamide Palmitoyl ethanolamide Linoleoyl ethanolamide
			Tryptophan metabolism	Serotonin	5-hydroxyindoleacetate N-formylanthranilic acid
			Sphingolipid metabolism		Sphingomyelin
			Fatty acid metabolism	Docosahexaenoylcarnitine
			Urea cycle	Prolylhydroxyproline
			Arginine and Proline metabolism	Prolylhydroxyproline
			Methionine, cysteine, and Taurine metabolism	Alpha-ketobutyrate
Smith et al. 2019
516 ASD compared with 164 typical children	1.5-4	Targeted approach using LC-MS	BCAA metabolism	Glutamine-to-isoleucine ratio Glutamine-to-leucine ratio Glutamine-to-valine ratio Glycine-to-isoleucine ratio Glycine-to-leucine ratio Glycine-to-valine ratio Ornithine-to-isoleucine Ornithine-to-leucine ratio Ornithine-to-valine ratio		There are abnormalities of BCAA metabolism in plasma of ASD.
Rangel-Huerta et al. 2019
30 ASD compared with 30 typical children	2-6	Untargeted approach using LC-MS	Neurotransmitter		Glutamate	There are abnormalities of neurotransmitter, reactive oxidative stress, BCAA, NAD, phospholipid metabolism in plasma of ASD.
			Reactive oxidative stress	Arginine N-acetylarginine Homoarginine
			BCAA metabolism	3-methyl-2-oxobutyrate 3-methyl-2-oxovalerate 4-methyl-2-oxopentanoate Isovalerylcarnitine Isobutyrylcarnitine
			NAD+ metabolism	1-methylnicotinamide N-methyl-2-pyridone-5-carboxamide	Nicotinamide
			Phospholipid metabolism		1-palmitoyl-glycerol-phosphatidyl-etholamine 1-stearoyl-glycerol-phosphatidyl-etholamine
			Others	n-alpha-acetylornithine Tryptophan Kynurenine 5-bromotryptophan 3-indoxyl sulfate Indole lactate 6-hydroxyindole sulfate
Orozco et al. 2019
167 ASD 51 i-DD 31 DS 193 typical children	2-5	Untargeted approach using 1H-NMR	One carbon metabolism	Glycine Serine		There are abnormalities of one carbon metabolism, TCA cycle, urea cycle in plasma of ASD.
			TCA cycle	Cis-aconitate
			Urea cycle	Ornithine
Lv et al. 2018
60 ASD compared with 30 typical children	2-5	Targeted approach using MS/MS	Fatty acid metabolism		Free carnitine Glutaryl carnitine Octyl carnitine Twenty-four carbonyl carnitine Carnosyl carnitine	There are abnormalities of fatty acid metabolism (free carnitine, short and long chain acyl carnitines) in plasma of ASD.
Anwar et al. 2018
38 ASD compared with 31 typical children	5-12	Targeted approach using LC-MS	Protein glycation	Nε-carboxymethyllysine Nω-carboxymethyllysine	3-deoxyglucoosone-derived hydroimidazolone	There are abnormalities of protein glycation and protein oxidation metabolism in plasma of ASD.
			Protein oxidation	Dityrosine
Wang et al. 2016
173 ASD compared with 163 typical children	3-6	Untargeted approach using UPLC/Q-TOF MS/MS	Fatty acid metabolism		Docosahexaenoic acid Docosapentaenoic acid	There are abnormalities of fatty acid, sphingomyelin, lysophosphatide, and beta oxidation in plasma of ASD.
			Sphingomyelin metabolism		Sphingosine-1-phosphate
			Lysophosphatide metabolism		LPA LysoPE
			Fatty acid beta oxidation	Decanoylcarnitine	9,10-Epoxyoctadecenoic acid
			Others	Pregnanetriol	Adrenic acid Uric acid
West et al. 2014
52 ASD compared with 30 typical children	4-6	Targeted approach using GC-MS and untargeted approach using LC-HRMS	TCA cycle	Succinate	Citrate	There are abnormalities of TCA cycle, fatty acid, oxidative phosphorylation, mitochondrial dysfunction, and gut microbiome metabolism in plasma of ASD.
			Fatty acid metabolism and oxidative stress		Methylhexa-decanoic acid Hepta-decanoic acid
			Oxidative phosphorylation	Aspartate Glutamate
			Mitochondrial energy production	DHEA-S	Isoleucine
			Gut microbiome		4-hydroxyphenyllactate
			Others	Glutaric acid 3-aminoisobutyric acid	Creatinine Homocitrulline
Kuwabara et al. 2013
25 ASD compared with 28 typical children	25-40 male	Untargeted approach using CE-TOF-MS	Oxidative stress	Arginine Taurine	5-Oxoproline	There are abnormalities of oxidative stress and mitochondrial metabolism in plasma of ASD.
			Mitochondrial dysfunction		Lactic acid
Delaye et al. 2018
22 ASD 29 ID 30 typical children	> 18	Targeted approach using amino acid chromatography	Neurotransmitter		Glutamate Serine	There are abnormalities of neurotransmitter metabolism in plasma of ASD.
			Others		Proline Ornithine

ASD autism spectrum disorder, i-DD idiopathic-developmental delay, DS Down syndrome, ID intellectual disability, ASQ Ages and Stages Questionnaire, UPLC-MS/MS ultra-performance liquid chromatography-tandem mass spectrometer, LC–MS liquid chromatography–mass spectrometry, GC–MS gas chromatography-mass spectrometry, ¹H-NMR hydrogen-1 nuclear magnetic resonance, MS/MS tandem mass spectrometry, UPLC/Q-TOF–MS ultra-high performance liquid chromatography-quadrupole time-of-flight mass spectrometry, LC-HRMS liquid chromatography-high resolution mass spectrometry, CE-TOF–MS capillary-electrophoresis-time-of-flight-mass-spectrometer, NAD nicotinamide adenine dinucleotide, LPA lysophosphatidic acid, LysoPE lysophosphatidylethanolamine, BCAA branched-chain amino acids, TCA tricarboxylic acid cycle, DHEA-S dehydroepiandrosterone sulfate

Alteration in the endocannabinoid metabolic pathway may be related to ASD. Endocannabinoids are neuromodulators that play a critical role in regulating emotional, behavioral, and social interaction. Kelly et al. (2019) found increased levels in specific endocannabinoids in children predicted to be at risk for ASD.

Disturbed neurotransmitter homeostasis has been proposed as possible pathogenesis of ASD. The decrease of glutamate level in the striatum was related to the symptom severity in autistic adult patients (Horder et al. 2018). Also, the perturbation of the malate-aspartate shuttle activity presented by the mitochondrial system might affect glutamate function (Monné et al. 2019). This was supported by the perturbation of glutamate-aspartate transporter activity in the gray matter of children with ASD (Palmieri et al. 2010). Additionally, two previous studies revealed decreased concentrations of glutamate in blood samples of children with ASD compared to age-matched controls (Delaye et al. 2018; Rangel-Huerta et al. 2019). On the other hand, some studies demonstrated increased blood levels of glutamate in cases of ASD (Cai et al. 2016; Hassan et al. 2013; Naushad et al. 2013; Tirouvanziam et al. 2012). These different findings might be due to the variation in the age range of the samples, study methodology and sample collection techniques, or the possible heterogeneity nature of the subjects.

ASD symptoms have been associated with pathways of oxidative damage (Chauhan and Chauhan 2006). This was suggested by the abnormalities in the nitric oxide pathway, as indicated by increased levels of arginine and its downstream metabolites in blood samples of children with ASD, especially those with developmental regression (Kuwabara et al. 2013; Rangel-Huerta et al. 2019). Moreover, an increase in blood taurine levels in ASD cases was observed (Kuwabara et al. 2013). Taurine is thought to be a neuroprotective compound against oxidative stress reactions (Niu et al. 2018), and therefore an increased concentration of taurine might reflect the compensation of oxidative stress.

Nicotinamide adenine dinucleotide (NAD) metabolism is also related to ASD, as indicated by the lower levels of nicotinamide in the plasma of children with ASD than that of typical children (Rangel-Huerta et al. 2019). NAD synthesis is important in several metabolic pathways, such as the TCA cycle, glycolysis, and fatty acid oxidation. The assumption is that decreased levels of nicotinamide might be due to the less preferential transformation of tryptophan to the melatonin pathway, leading to an accumulation of oxidative stress (Mussap et al. 2016).

Lipid metabolism complications have been assumed to be a causative mechanism of ASD. Metabolomic profiles in the blood revealed decreased levels of lysolipids (Rangel-Huerta et al. 2019), free carnitine (Lv et al. 2018), short- and long-chain acylcarnitines (Lv et al. 2018), and docosahexaenoic acid (DHA) (Wang et al. 2016) in children with ASD in comparison to age-matched controls. Lower levels of polyunsaturated fatty acids that play a vital role in the structural and functional integrity of neurons were found in children with ASD (Mazahery et al. 2017; Parletta et al. 2016). The addition of DHA as a supplement could potentially improve behavior in children with ASD, adding weight to this finding (Amminger et al. 2007; Yui et al. 2012). In addition, abnormal white matter development has been proposed as being related to abnormal sphingomyelin metabolism in ASD, as indicated by increased levels of sphingosine-1-phosphate in the blood of children with ASD (Wang et al. 2016).

The link of Urea cycle defects to ASD was supported by the evidence of higher ornithine levels in blood samples of children with ASD (Orozco et al. 2019), which might result from ornithine transcarbamylase (OTC) dysfunction. The evidence of OTC deficiency related to ASD-like symptoms was presented in a case report (Görker and Tüzün 2005).

Disruption of the TCA cycle has been reported as being associated with ASD. This was evidenced by the increased blood levels of cis-aconitate (Orozco et al. 2019) and succinic acid (West et al. 2014), but decreased blood levels of citric acid (West et al. 2014) in children with ASD compared to children with typical development.

Proteotoxic stress induced by protein glycation and oxidation may be linked to the pathogenesis of ASD. The elevation of advanced glycation end products (AGEs) and dityrosine residue were revealed in the plasma of children with ASD (Anwar et al. 2018). It is proposed that the accumulation of AGEs is associated with perturbation of the lipid peroxidation process, and the increased levels of markers of this process, such as hexanoyl-lysine adduct, urinary 8-isoprostane-F2a, and plasma malondialdehyde, have also been related to ASD symptoms (Chauhan and Chauhan 2006; Ghezzo et al. 2013; Ming et al. 2005). The increased level of dityrosine was a result of increased dual oxidase (DUOX) activity, which was related to gut mucosal dysfunction (Bae et al. 2010; Chang et al. 2013).

The disruption of oxidative phosphorylation in mitochondria could also be related to ASD. The higher levels of aspartate and glutamate in blood samples of children with ASD may be linked the aspartate/glutamate transporter gene mutation, SLC25A12 (West et al. 2014). SLC25A12 plays a crucial role in the malate/aspartate shuttle producing cellular energy via oxidative phosphorylation (West et al. 2014). The disruption of the respiratory chain reaction was also indicated by the increased levels of dehydroepiandrosterone (DHEAS) in children with ASD (West et al. 2014). DHEAS can inhibit the enzymes related to the mitochondrial respiratory chain process, thus interfering with energy production (Safiulina et al. 2006).

Alteration in mammalian-microbial co-metabolism has also been considered as being a mechanism involved in ASD pathogenesis. Decreasing levels of 4-hydroxyphenyllactate were observed in the blood of children with ASD compared to typical children (West et al. 2014). This metabolite is related to bacteria: lactobacilli and bifidobacteria, which act as antioxidants (Beloborodova et al. 2012). This finding was supported by decreased antioxidant enzymes such as superoxide dismutase and glutathione peroxidase in the blood of children with ASD (Meguid et al. 2011; Yorbik et al. 2002).

The alteration of metabolomes in the blood samples could reflect the several abnormal metabolic pathways associated with children with ASD. The metabolism of some amino acids, such as tryptophan and BCAA, and neurotransmitters, such as glutamate, reactive oxygen species, and lipids, might play an important role in ASD pathogenesis. Further investigation on novel biomarkers from those abnormal metabolic pathways was needed.

Pharmacological interventions for ASD: metabolomic studies in animals

The research into pharmacological interventions for ASD from metabolomic studies in animals is limited, as shown in Table 3. The intervention in animal studies is suramin, an antipurinergic agent. Naviaux et al. (2014) gave suramin to mice from maternal immune activation (MIA) models. The concept of MIA model was to make the offspring mice having the features of ASD by letting them exposed to the poly(Inosine:Cytosine), which was intraperitoneally injected to the pregnant dams. Exposure to poly(Inosine:Cytosine) could lead to cell danger response (CDR), which is considered a metabolic response mechanism to a threat (Naviaux 2014). They found the near-normalization of blood metabolomic profiles in treated mice had improved social and novelty preference testing. (Naviaux et al. 2014). The metabolomics profiles alterations included increased levels of adenosine triphosphate and allantoin, whereas decreased levels of adenosine monophosphate, inosine monophosphate, deoxyadenosine monophosphate, deoxyinosine monophosphate, guanosine, inosine triphosphate, and guanosine diphosphate. The purinergic signaling regulates the CDR. When there was a perturbation in purine metabolism, it causes the persistence of CDR which leads to several chronic diseases such as ASD (Naviaux 2014). Hence, antipurinergic therapy has been thought to correct the dysregulation of purine metabolism. Antipurinergic therapy, in addition to mitigate autistic-like symptoms in ASD caused by the environmental factor (MIA model), could improve such symptoms in ASD caused by a genetic factor such as Fragile X (Fmr1 knockout) mouse model. Fragile X syndrome is the most common known inherited single-gene disorder associated to ASD (Cogram et al. 2020). The study of Naviaux et al. (2015) emphasized the association of ASD features and perturbed purinergic signaling pathway by showing the normalization of purine metabolism-related metabolites and restoration of typical social behaviors in Fmr1 knockout mice treated by antipurinergic therapy (Naviaux et al. 2015).

Table 3.

Pharmacological interventions for autism spectrum disorder: metabolomic studies in animals

Ref and Model	Age	Intervention Dose/ duration	Method	Metabolic pathway	Changes in metabolites compared to controls		Outcomes	Interpretation
					Increase	Decrease
Naviaux et al. 2014
Male mice from MIA model of ASD features vs. controls	6 mo	Suramin 20 mg/kg intraperitonal single dose vs. NSS	Targeted approach using LC-MS in blood	Purine metabolism	Inosine	Adenosine monophosphate	Improvement in social and novelty preferences testing	Single dose of suramin could reverse social deficit in MIA model by normalization of metabolites in many metabolic pathways, mainly in purine metabolism.
					Deoxyinosine	Inosine monophosphate
					Adenosine triphosphate	Deoxyadenosine monophosphate
					Allantoin	Deoxyinosine monophosphate Guanosine Inosine triphosphate Guanosine diphosphate
				Microbiome metabolism	Benzoic acid	Hydroxyphenylacetic acid
					Histidinol	Sucrose-6-phosphate
					dTDP-glucose	3-Hydroxyanthranilic acid
Naviaux et al. 2015
Fragile X (Fmr1) knockout mouse model vs. controls	9 wk	Suramin 20 mg/kg intraperitonially weekly 16 wk vs. NSS	Targeted approach using LC-MS/MS in blood	Purine metabolism		Xanthine Hypoxanthine Inosine Guanosine	Restoration of social preference, social novelty, and spontaneous altered behavior	Suramin could restore typical social behaviors in Fmr1 knockout mouse model by normalization of metabolites in many metabolic pathways, mainly in purine metabolism.
				Fatty acid oxidation		Hexanoylcarnitine 3,5-Tetradecadiencarnitine Myristoylcarnitine
				Eicosanoid metabolism	Leukotriene B4 Epoxy-5,8,11-eicosatriencic acid 8,9-Epoxyeicosatrienoic acid	11-Dehydrothromboxane B2

LC–MS liquid chromatography-mass spectrometry, LC–MS/MS liquid chromatography–tandem mass spectrometry, APT antipurinergic therapy, RTT Rett syndrome, NSS normal saline solution, MIA model maternal immune activation model, mo months, wk weeks

In summary, the metabolomic studies using pharmacologic intervention for ASD demonstrated abnormalities in various metabolic pathways, as shown in Fig. 1, including purine metabolism, microbiome, fatty acid oxidation, TCA cycle, and eicosanoid metabolism. Near-normalization of the metabolite levels following those pharmacological interventions, such as suramin, was associated with a near-restoration of social behaviors in mice with autistic features.

Fig. 1.

The altered metabolic pathways from the combination of pharmacologic interventions in Tables 3 and 4 were shared by animals with ASD features and humans with ASD. Four metabolic pathways were shared by animals with ASD features. Seventeen metabolic pathways were shared by humans with ASD after therapeutic trials. The potential common pathways shared by both animal and human studies were mammalian-microbial co-metabolite, purine metabolism, and fatty acid oxidation

Pharmacological interventions for ASD: metabolomic studies in humans

Details of the metabolomic studies of pharmacological interventions for ASD in humans are shown in Table 4. Those interventions include sulforaphane, and prebiotic: bimuno galactooligosaccharide (B-GOS).

Table 4.

Pharmacological interventions for autism spectrum disorder: metabolomic studies in humans

Ref and Model	Age	Intervention Dose/duration	Method	Categories of Metabolites/ Involved metabolic pathways	Changes in metabolites		Clinical Outcomes	Interpretation
					Increase	Decrease
Bent et al. 2018
15 ASD children before and after intervention	5-22 y	Sulforaphane 2.5 μmol glucoraphanin per lb/12 wk	Untargeted approach using UHPLC-MS/MS in Urine	Arginine	N-acetylputrescine		Improvement of behavior (ABC) and social (SRS) responsiveness score	Sulforaphane supplementation statistically significantly improves behavior and social responsiveness by changes in urine metabolites involved in several pathways including amino acid, cholesterol, neurotransmitter, oxidative stress, polyol associated with uremia, sphingomyelin, TCA cycle, and phospholipid metabolism.
				Tryptophan	Tryptophan
				Tyrosine	Tyrosine
				Leucine metabolism	β-hydroxyisovalerate α-hydroxyisocaproate
				Isoleucine	α-hydroxyisovalerate
				Valine	3-hydroxyisobutyrate
				Cysteine	Taurine
				Mammalian-microbial co-metabolism	Tryptophan betaine 4-Hydroxybenzoate 3-Ethylphenylsulfate Arabinose Erythriol
				Cholesterol	Cholesterol Cholate 12-Dehydrocholate
				Cholesterol (bile acid), microbiome origin	Glycocholenate sulfate
				Cholesterol (hormone)	Cortisone Cortisol 21-glucuronide Epiandrosterone glucuronide 17α-hydroxypregnanolone gluconide 5α-androstan-3β,17α-diol disulfate Pregnen-diol disulfate Dehydroepiandrosterone 11-Ketoetiocholanolone sulfate 5α-pregnan-3β,20α-diol disulfate 5α-pregnan-3(α/β),20β-diol disulfate 21-Hydroxypregnenolone disulfate
				Neurotransmitter	Glutamine Serotonin Homovanilate Hypoxanthine
				Oxidative stress	γ-Glutamylglutamine Methionine sulfone
				Polyol associated with uremia	3-Carboxy-4-methyl-5-propyl-2-furanpropanoate
				Sphingomyelin	Stearoyl sphingomyelin Lignoceroyl sphingomyelin Behenoyl sphingomyelin Sphingomyelin
				TCA cycle	Malate
				Phospholipid	1-Palmitoyl-2-oleoyl--sn-glycero-3-phosphoethanolamine
Grimaldi et al. 2018
30 ASD on exclusion or unrestricted dwiet with intervention vs. placebo	4-11 y	B-GOS® 6 wk vs. placebo	Targeted approach using 1H-NMR in Feces and Urine	Amino acid	Dimethylglycine (urine)	Phenylacetylglycine (urine)	Improvement of antisociability and social skill scores	Prebiotics (B-GOS) intervention, especially with exclusion diet, help improvements in antisocial behaviors in ASD by modulating gut microbiota and changing in fecal and urine metabolites.
					Dimethylglycine (feces)	Phenylalanine (urine)
					Dimethylalanine (urine)	Isoleucine (feces) Leucine (feces) Valine (feces) Alanine (feces) Glutamine (feces)
				Fatty acid	Adipate (urine) Butyrate (feces) Valerate (feces)
				Organic acids	Citrate (urine)	Lactate (feces) B-hydroxybutyrate (urine)
				Others	Creatinine (urine) Creatine (urine) Carnitine (urine) Trimethylamine-N-oxide (urine) Ethanol (feces)
Naviaux et al. 2017
10 ASD randomized to receive either suramin vs. placebo	4-17 y	Suramin 20 mg/kg single dose VS saline infusion	Targeted LC-MS/MS Blood	Microbiome	2-Keto-L-gluconate P-Hydroxyphenylacetic acid 4-Hydroxyphenyllactic acid		ADOS-2 comparison scores improved in suramin group but did not change in placebo group	Low-dose suramin was associated with improvement of ASD symptoms measured by ADOS-2.
				Sphingomyelin	Sphingomyelin(d18:1/26:0 OH)
				1-carbon, folate, glycine, serine	Glycine
				Purines	1-Methyladenine	cAMP
					Purine Allantoin	dGDP
				Amino acids	Alanine L-Asparagine
				Pyrimidines	Cytosine
				Krebs cycle	Citric acid Cis-aconitic acid
				GABA, Glutamate, Arginine, Ornithine, Proline	1-Pyrroline-5-carboxylic acid
				Dipeptides	Gamma-glutamyl-Alanine
				Histidine, histamine	Histamine
				Nitric oxide, ROS	Azelaic acid
				SAM, SAH, Met, Cys, GSH	Methionine sulfoxide Cysteamine S-Adenosylhomocysteine
				Tryptophan, Kyurenine	L-Kynurenine
				Glycolysis	Glycerol 3-phosphate
				Bile acids	Chenodeoxyglycocholic acid Glycocholic acid
				Vitamin C	Hydroxyproline
				Branch chain amino acids	2-Hydroxyisovaleric acid Isovalerylglycine Tiglylglycine
				Disaccharides	Hexose Disaccharide pool
				Tyrosine, phenylalanine	L-Phenylalanine Cinnamoylglycine
				Cholesterols	Lathosterol
				Fatty acid oxidation		Octanoylcarnitine

ASD autism spectrum disorder, ADOS-2 The Autism Diagnostic Observation Schedule-Second Edition, B-GOS® Bimuno® galactooligosaccharide, LC–MS liquid chromatography-mass spectrometry, LC–MS/MS liquid chromatography–tandem mass spectrometry, ¹H-NMR hydrogen-1 nuclear magnetic resonance, UHPLC-MS/MS ultra-high performance liquid chromatography-tandem mass spectrometry, SM sphingomyelin, cAMP cyclic adenosine monophosphate, dGDP 2’-deoxy-guanosine diphosphate, GABA Gamma-aminobutyric acid, SAM S-adenosylmethionine, SAH S-adenosylhomocysterin, Met methionine, Cys cysteine, GSH glutathione, TCA tricarboxylic acid cycle, ABC Aberrant Behavior Checklist, SRS Social Responsiveness Scale, y year, wk weeks

Sulforaphane

There has been a growing interest in sulforaphane, a substance extracted from broccoli sprouts, as a possible treatment option for ASD patients (McGuinness and Kim 2020). Sulforaphane is known for its proposed mechanism to reverse the perturbed metabolic pathways associating with ASD, including mitochondrial dysfunction, neuroinflammation, and oxidative stress (Singh et al. 2014). Bent et al. (2018) indicated an association between the changes in urine metabolites and improvement of ASD symptoms after treatment with sulforaphane in children and young adults with ASD (Bent et al. 2018). They found increased antioxidant-related substances such as γ-glutamylglutamine and methionine sulfone in the urinary after sulforaphane ingestion (Bent et al. 2018). They also found a strong correlation between the increased urinary levels of sphingomyelin and the improvement of ASD behaviors after treatment with sulforaphane. This was consistent with a previous study that demonstrated decreased levels of sphingomyelin in urine samples of children with ASD compared to age-matched controls (Kelly et al. 2019; Wang et al. 2016). Hence, restoration of the sphingomyelin metabolism may be a potential intervention for the improvement of ASD symptoms. Apart from its capacities of anti-inflammation and antioxidant, sulforaphane repairs immune dysfunction through activation of nuclear factor erythroid 2-related factor 2 (Nrf2). This mechanism mitigates ASD-like symptoms in the animal study (Nadeem et al. 2019).

B-GOS

Gut dysbiosis is one of the most likely causes of ASD pathogenesis. The possible link between alteration in gut microbiota and ASD symptom improvement has been investigated (Fattorusso et al. 2019; Kang et al. 2019; Li et al. 2017). Grimaldi et al. (2018) studied the metabolomic alterations in feces and urine between the groups receiving the prebiotic B-GOS and the controls taking a placebo of children with ASD either with exclusion or unrestricted diet (Grimaldi et al. 2018). They showed various changes in metabolites, including amino acids, fatty acids, and organic acids in the case of the ASD cohort with the prebiotic compared to those with placebo (Grimaldi et al. 2018). Interestingly, after B-GOS ingestion, the team observed the increased levels of butyrate in feces. Butyrate can be produced by bacteria in the Lachnospiraceae family, which is found in increased abundance after modulation of the composition of the gut microflora following B-GOS administration. In addition, decreased levels of some amino acids in children with ASD after treatment with B-GOS compared to those with placebo treatment were thought to be an effect of such prebiotic causing the increase in the absorption of amino acids through the intestine. (Grimaldi et al. 2018).

Suramin

An antipurinergic agent was shown to improve social behaviors in MIA and Fmr1 knockout mouse models by normalizing metabolites mainly in purine metabolism as presented in the previous animal studies (Naviaux et al. 2014, 2015). An explanation of the mechanism was discussed in the previous animal study section. For a human study, Naviaux et al. (2017) conducted a trial of single low dose suramin to children with ASD (Naviaux et al. 2017). They found the improvement of ASD symptoms associated with the alteration of several metabolic pathways. Notably, more than half of the pathways changed by suramin in children with ASD were also changed in the MIA mouse model. Therefore, suramin might be a promising agent for the treatment of ASD.

Due to limited trials on humans, we cannot conclude the strong evidence of the links between improved clinical symptoms of ASD and associated metabolic pathways. However, the potential intervention, such as sulforaphane, and B-GOS, may be beneficial for social behaviors in children with ASD by altering the perturbed ASD-related metabolites.

Conclusion and clinical implications

Previous metabolomic studies in urine and blood samples revealed that there were abnormalities of metabolism in children with ASD. These include the TCA cycle; urea cycle; oxidative stress; amino acids such as methionine, tryptophan, tyrosine, and BCAA; imbalance of neurotransmitters; mammalian-microbial co-metabolism; metabolic pathway of nucleotides, nicotinate and nicotinamide, lipids and fatty acids, carbohydrates, endocannabinoid, NAD; pentose phosphate pathway; protein glycation and oxidation, and oxidative phosphorylation. In animal studies, the pharmacological intervention that could improve ASD features by altering the metabolomic profiles was suramin. In human studies, the potential interventions that improved social behaviors in children with ASD include sulforaphane, and prebiotic B-GOS. The majority of studies that reported changes in metabolomic profiles after the treatment also demonstrated a favorable outcome of ASD symptoms. Several previous studies showed there are another promising pharmacological supplements for improving symptoms of ASD such as omega-3 fatty acid, folinic acid, and vitamin D. However, there is still a lack of evaluation in the metabolomics approach.

The most repeatedly proposed mechanisms and possible underlying pathophysiology of ASD were alterations of amino acids, reactive oxidative stress, neurotransmitters, and microbiota-gut-brain axis. For the interventional studies, it was less likely to explicitly present the trend of the most related metabolic pathways in relation to ASD, since only a few studies with various types of pharmacological interventions have been reported. However, our integrated metabolome data from both animal and human interventional studies reveal, to some extent, the shared potential pathways associated with the ASD phenotypic improvement. Those potential pathways were mammalian-microbial co-metabolite, purine metabolism, and fatty acid oxidation. Interestingly, all three interventional studies in humans also showed an increase in sphingomyelin, phospholipid-, and cholesterol-related metabolites, which were related to the improvement of ASD symptoms. This growing evidence of lipid alterations suggests that lipidomics might be another interesting field of investigation for a comprehensive explanation of ASD development in the future (El-Ansary et al. 2020).

Currently, the profiles of metabolic changes in previous studies have been inconsistent. Most notably, the ethnicity of the participants, which causes differences in both genetic and environmental exposure, may be the key factor in explaining unpredictable results. Much of the variation could be due to the differing populations, the types of samples, sample collection and sample preparation techniques, and data analysis techniques. Moreover, the varying degrees of symptom severity, the difference of co-morbidities, and various types of medications could influence the variety of metabolomic profiles among patients. The subgroup analysis investigating the association of those variations and metabolic profiles should be more researched. Prior studies opened the question of whether the heterogeneity, for example – symptom severity and age, could affect the change in metabolomic profiles among individuals (Mussap et al. 2020; Olesova et al. 2020). Mussap et al. (2020) showed the correlation between phenotypic severity of autism and urine metabolomes. Indeed, they found that the levels of some metabolites derived from diet, gut dysbiosis, tryptophan metabolism, and mitochondrial dysfunction of autistic children with severe core symptoms were distinguishable from those with less severe core symptoms (Mussap et al. 2020). In addition, Olesova et al. (2020) indicated that there was a significant difference in metabolomic profiles between ASD and non-ASD groups only in the school-aged children but not in the preschoolers (Olesova et al. 2020).

For further studies, confounding factors such as dietary intake, medication use, and illness in both children with ASD and children without ASD should be controlled. The severity of ASD, comorbid diseases, ethnicity, cognitive level, and environmental factors should also be collected for subgroup analysis to explain the pathogenesis of ASD more extensively. More metabolomic studies should focus on the pharmacological treatments and correlation of metabolomic profiles and ASD phenotypic changes after treatments. Importantly, there is a need for evaluation of the model sensitivity and specificity and cost-effectiveness for the practical use.