Exploring Causal Associations Between Plasma Metabolites and Autism Spectrum Disorder

Abstract

Background:

In autism spectrum disorder (ASD), the human plasma metabolome is altered but the causal relationship between the levels of metabolites and ASD is unclear. We aimed to assess bidirectional causal associations between plasma metabolites and ASD.

Methods:

We investigated potential causal associations between the genetic variation contributing to the levels of metabolites and ASD via Mendelian randomization (MR) analyses. Genome-wide association study (GWAS) summary datasets were used in the study, including ASD (n = 46,350) and 871 plasma metabolite (n = 8299) datasets. We used druggability analysis to prioritize metabolites with therapeutic potential.

Results:

Our MR analysis identified 32 plasma metabolites whose levels were protective against the risk of ASD, including 5 alpha-androstan-3 alpha, 17 beta-diol disulfate (odds ratio (OR): 0.94, 95% CI: 0.90–0.97) and 11beta-hydroxyetiocholanolone glucuronide (OR: 0.95, 95% CI: 0.92–0.98). Additionally, 12 metabolites were found to be positively associated with the risk of ASD, including indoleacetylglutamine (OR: 1.04, 95% CI: 1.01–1.08) and sphingomyelin (d18:1/24:1, d18:2/24:0) (OR: 1.06, 95% CI: 1.01–1.11). Some metabolites may be regulated through drug intervention, including sphingomyelin, chiro-inositol, carotene diol (1)/(2), and glycerol. Genetic variation contributing to ASD may increase the abundance of five metabolites, including deoxycholic acid glucuronide (OR: 1.18, 95% CI: 1.03–1.34); meanwhile, the abundance of 27 metabolites, including stearoyl choline (OR: 0.80, 95% CI: 0.69–0.92) may be causally reduced.

Conclusions:

Our MR analysis uncovered bidirectional causal associations between certain plasma metabolites and ASD, suggesting that these metabolites could be biomarkers for ASD and paving the way for novel therapeutic targets in ASD phenotypes.

Graphical abstract

Keywords

autism spectrum disorder / plasma metabolite / Mendelian randomization / causal association

Main Points

1. In autism spectrum disorder (ASD), alterations occur in the human plasma metabolome. However, the causal relationship between metabolite levels and ASD remains ambiguous.

2. Mendelian randomization (MR) studies that utilize genetic variants as instrumental variables are now widely used to infer causality from a genetic perspective.

3. Our research indicates that specific plasma metabolites play a role in increasing the risk of ASD. Conversely, ASD has the potential to influence the composition of metabolites.

1. Introduction

Autism spectrum disorder (ASD) is a lifelong neurodevelopmental disorder arising from a constellation of genetic and environmental factors, with a heritability rate of approximately 85%. It is characterized by severe impairment of social interaction and communication skills, accompanied by restricted, repetitive, and stereotyped behavior patterns and interests [1, 2]. The prevalence rate of ASD is approximately 1.5% and it is increasing annually, imposing a heavy economic burden on families and society [3].

Patients with ASD often present with various other conditions such as anxiety, attention-deficit/hyperactivity disorder, and depression, which increases the complexity of ASD diagnosis and treatment. Due to the lack of specific biomarkers or imaging tests, diagnosis of ASD relies on clinical assessment, behavioral observations, and standardized questionnaires. Clinical manifestations of ASD exhibit significant heterogeneity, thus posing additional problems [4].

ASD patients exhibit overall decreased cellular energy balance and insufficient mitochondrial energy reserves, which may contribute to cognitive impairments, language deficits, and energy metabolism abnormalities. ASD-associated mitochondrial dysfunction may lead to reduced synaptic neurotransmitter release, which is observed in GABAergic interneurons. Increased susceptibility to oxidative stress in ASD patients may be linked to altered antioxidant enzyme activity, resulting in mitochondrial dysfunction and impaired energy metabolism [5, 6]. Mitochondrial dysfunction may not only be comorbid with ASD but also potentially a contributing factor [7].

Metabolites are small molecules formed during biochemical reactions. These molecules may be either intermediates or end products such as amino acids, nucleotides, lipids, hormones, and exogenous substances participating in human metabolism. Levels of individual metabolites are influenced by inherited genetic variants, diet, and other environmental exposures [8]. Novel metabolomics platforms continue to emerge and their application is helping to both uncover the pathogenesis of human diseases and identify novel biomarkers for diagnoses and prognoses [9, 10].

Alterations in plasma or serum metabolite levels have been found in previous studies of neuropsychiatric conditions [11, 12]. Of note, Zhang et al. [13] identified an association between the levels of 38 plasma metabolites and an increased risk of dementia. A large-scale study involving 5283 patients with major depressive disorder (MDD) and 10,145 controls showed that the levels of 21 metabolites were significantly associated with depression [14]. Similarly, when the blood and urine samples of ASD patients were compared with those collected from normally developing children, significant differences in the abundance of various metabolites were uncovered. The potential for distinguishing individuals with ASD from normal children was independently identified for compounds belonging to a variety of metabolite classes, including amino acids, lipids, and xenobiotics [7, 15, 16].

For many metabolites, their serum levels are highly heritable [17, 18]. Understanding if their involvement in ASD is causal may be beneficial for understanding the pathogenesis of this disease and the development of potential strategies for its treatment.

To evaluate the causal effects of each risk factor on a certain outcome, Mendelian randomization (MR) employs genetic variants as instrumental variables. Like randomized assignment of individuals to the treatments in clinical trials, the random assortment of genetic variants in an individual’s diploid genome provides a framework to evaluate the causality of the observed associations between exposures and outcomes. These associations are inferred from aggregated data collected in genome-wide association studies (GWAS). This approach minimizes the effects of confounding factors, including age, other drug or environmental exposures, and reverse causation [19, 20], and now is widely used to infer causality from a genetic perspective [21, 22, 23, 24, 25]. In this study, we used MR analysis to explore the causal effects connecting various small molecular constituents of plasma and ASD. We then evaluated whether the identified ASD-related metabolites can be modulated by pharmacological or other interventions.

2. Methods

2.1 Data Sources and Study Design

The GWAS summary results used for this analysis were all derived from publicly available data. The GWAS data for ASD (n = 46,350; 18,381 cases and 27,969 controls) were sourced from the Integrative Psychiatric Research (iPSYCH) and Psychiatric Genomics Consortium (PGC). The PGC samples were retrieved from five European cohorts. The iPSYCH samples were gathered from a population-based cohort consisting of all children born in Denmark between 1981 and 2005 [26].

For metabolites, we focused on data from the Canadian Longitudinal Study of Aging (CLSA). The research targeted 8299 unrelated European subjects within the CLSA, all of whom underwent genome-wide genotyping and blood metabolite measurement. The CLSA tracks over 50,000 Canadians aged 45–85 years (50.9% female) for various types of information such as biological and medical data. The levels of blood metabolites were measured using the ultrahigh-performance liquid chromatography-tandem mass spectrometry (UPLC-MS/MS) platform by Metabolon Inc. (Morrisville, CA, USA) [27].

Among the plasma metabolites tested, those with known identities spanned eight super-pathways (lipids, amino acids, xenobiotics, nucleotides, cofactors and vitamins, carbohydrates, peptides, and energy). After excluding the metabolites labeled as “X-” (unknown), we tested 871 metabolites. When analyzing the metabolic pathways, we referred to the original research literature, which contained the metabolic pathways of these metabolites. Ethical approval was obtained in all original studies. A flowchart outlining the current study is shown in Supplementary Fig. 1.

2.2 Instrumental Variables (IVs)

We selected candidate instrumental variables (IVs) based on a genome-wide significance threshold of p 1.0 \times 10^-5. Single nucleotide polymorphisms (SNPs) were selected using the 1000 Genomes Project Phase 3 (EUR) reference panel [28] and, to ensure the independence of the IVs, we performed pruning with an r² threshold of 0.001 within a 10 Mb window. This strategy helped to ensure that the IVs included in the analysis were independent, thereby improving the precision and robustness of the MR results.

2.3 MR Analysis

In this study, we utilized three different models from the TwoSampleMR package (https://mrcieu.github.io/TwoSampleMR) in the R software (R Core Team, Vienna, Austria, version 4.0.5) to investigate causal relationships [29]. We primarily utilized the inverse-variance-weighting (IVW) model, which assumes a zero intercept and estimates causal effects through fixed-effect meta-analysis. To ensure the robustness of our findings, we conducted sensitivity analyses using additional MR methods, such as the weighted median (WM) and MR-Egger methods, which differ from IVW in their assumptions. The latter model, for instance, assumes that pleiotropic effects are independent and employs weighted linear regression of outcome coefficients against exposure coefficients. We assessed the presence of pleiotropy by examining the intercepts of MR-Egger regression, which provided insights into mean-level pleiotropy. Furthermore, we evaluated the heterogeneity of our results using both I² statistics and Cochran’s Q test, considering values where both I² > 0.25 and P_Q 0.05 are indicative of significant heterogeneity. An IVW-based p 0.05 indicated a significant correlation between exposure and outcome.

2.4 Druggability Evaluation

We searched for targets and drug information in DrugBank [30] and ChEMBL [31] to assess whether the identified metabolites could serve as potential therapeutic targets. These databases prioritized druggable targets by integrating information from text mining, gene function, drug-gene interactions, and expert curation.

3. Results

3.1 Selection of Instrumental Variables

Following an extensive quality control procedure, we discovered a total of 18,669 SNPs associated with metabolites as IVs for investigating the association between metabolites and ASD. In the reverse MR analysis, we included a total of 54 IVs associated with ASD. Detailed descriptors of these SNPs are shown in Supplementary Tables 1,2.

3.2 MR Analysis

Using IVW, a total of 44 metabolites were identified as conferring causal effects on ASD (Table 1 and Fig. 1A). Of these, 12 were found to be significantly and positively associated with the risk of ASD, including indoleacetylglutamine and sphingomyelin (d18:1/24:1, d18:2/24:0). Conversely, 32 metabolites, including 5alpha-androstan-3alpha,17beta-diol disulfate and 11beta-hydroxyetiocholanolone glucuronide, were highlighted as protective factors against ASD (positive effects shown in red and negative effects shown in blue in Fig. 2A–C).

Reverse MR analysis revealed that genetic liability to ASD is linked to increased levels of five metabolites, including deoxycholic acid glucuronide and stearidonate (18:4n3), as well as a reduction in 27 metabolites such as stearoyl choline, 4-methylcatechol sulfate, and taurine (Table 2; Fig. 1B; positive effects shown in red and negative effects shown in blue in Fig. 2B,C).

The results showed that metabolites were mostly concentrated at the lipid and amino acid levels, and the metabolic pathways involved included androgenic steroid biosynthesis, tryptophan metabolism, primary bile acid metabolism, creatine metabolism, and purine metabolism. For details on other relevant metabolic pathways, see Supplementary Table 3.

Except for adenosine 5^′-diphosphate (P_Q 0.05), Cochrane’s Q-test indicated that there was no significant heterogeneity between most SNPs. However, the MR-Egger model analysis detected some horizontal pleiotropy involving adenosine 5^′- diphosphate, glucuronide of C₁₀H₁₈O₂ (7), and deoxycholic acid glucuronide (p_pleiotropy 0.05), amongst others. Therefore, these results should be interpreted with caution (Supplementary Tables 4,5).

3.3 Five Metabolites as Therapeutic Drug Targets

Druggability evaluation showed that five of the ASD-related metabolites (sphingomyelin, chiro-inositol, carotene diol (1)/(2), and glycerol) have been targeted by pharmacologic intervention. One of these drugs, olipudase alfa, that targets sphingomyelin (d18:1/24:1, d18:2/24:0) has been used to treat acid sphingomyelinase deficiency by catalyzing the hydrolysis of sphingomyelin, thereby reducing its accumulation. Inositol and beta-carotene are used as nutritional supplements in special dietary foods and infant formula. Inositol plays an important role in ensuring oocyte fertility, and its association with D-chiro inositol has also been studied in polycystic ovary syndrome. The use of glycerin may help to improve gastrointestinal symptoms in ASD patients (Supplementary Table 6).

4. Discussion

In this study, we explored potential associations between human plasma metabolites and ASD by conducting a two-sample MR analysis on GWAS summary statistics datasets. We identified 44 metabolites with underlying genetic variation causally related to ASD risk. Druggability evaluation prioritized five ASD-related metabolites (sphingomyelin, chiro-inositol, carotene diol (1)/(2), and glycerol) that could be modified by drug interventions. In addition, the genetic component of ASD affected the levels of 32 metabolites.

Previous research has shown that there are significant differences in the metabolic profiles between children with ASD and neurotypical children. These differences extensively involve various metabolic pathways that involve lipids, amino acids, nucleotides, and energy. These differences may change with age [15, 32, 33]. Liu et al. [34] conducted a quantitative analysis of 28 children with ASD and identified three potential biomarkers. The findings of our study demonstrate that the metabolites are primarily related to lipids and amino acids. The metabolic pathways involved include androgenic steroid biosynthesis, tryptophan metabolism, primary bile acid metabolism, creatine metabolism, purine metabolism, methionine, cysteine, S-Adenosylmethionine (SAM), and taurine metabolism.

An imbalance in oxidative stress responses and pro-inflammatory processes [35] is a commonly observed feature in patients with ASD, who often exhibit metabolic abnormalities associated with mitochondrial dysfunction, affecting cellular energy production [36, 37, 38, 39]. The tricarboxylic acid (TCA) cycle occurs in the mitochondrial matrix and is crucial for many cellular processes. Our research indicates that the genetic signature contributing to the elevation of the levels of cis-aconitate, a TCA intermediate, raises the risk of ASD. Sotelo-Orozco et al. [40] demonstrated that an increase in cis-aconitate levels is linked to poorer cognitive and adaptive skills in children with ASD. Likhitweerawong et al. [16] also confirmed that cis-aconitate was significantly elevated in patients with ASD.

Moreover, the activity of aconitase, an enzyme that degrades cis-aconitate, is lower in the autism cerebellum and is negatively correlated with the reduced glutathione/oxidized glutathione (GSH/GSSG) ratio, which indicates glutathione redox/antioxidant capacity [41]. GSH plays a protective role against oxidative stress and neuroinflammation. Depletion of GSH following chronic gastrointestinal issues, which are common in ASD patients, is also associated with mitochondrial dysfunction [42, 43]. Thus, it is not surprising that our reverse MR analysis showed that the lowering of the levels of 5-oxoproline, a major intermediate in the \gamma-glutamyl cycle whereby glutathione is produced and then broken down, is also associated with ASD occurrence.

A study involving 20 Egyptian children aged 2–7 years found that the severity of autism symptoms, measured by the Childhood Autism Rating Scale (CARS), negatively correlates with levels of leucine, isoleucine, phenylalanine, cysteine, serine, and tyrosine [40]. Notably, the absence of the solute carrier transporter 7a5 (SLC7A5) at the blood-brain barrier leads to a significant reduction in the levels of large neutral amino acids (LNAAs) in the brain parenchyma, possibly contributing to the pathogenesis of ASD [44]. Moreover, the metabolism of LNAAs and lipids are interconnected, with the deletion of SLC7A5 in the neurons affecting their metabolic state, leading to a shift in lipid metabolism [45].

Glutamatergic dysfunction has been implicated in the pathophysiology of ASD, with multiple studies reporting abnormalities in glutamate levels across different regions of the ASD brain [46, 47]. Our reverse MR analysis indicated that a reduction in S-1-pyrroline-5-carboxylate, involved in glutamate metabolism, is associated with the risk of ASD. On the other hand, the genetic signature predisposing to ASD is associated with a decrease in the levels of taurine, a gliotransmitter and an antioxidant. A previous study connected deficiencies in hypotaurine and taurine production with defects in cell differentiation in the brain [48]. ASD-related taurine pathway abnormalities may further impact neuronal signaling and perpetuate a vicious metabolic circle supporting the pathophysiology of ASD [49, 50]. Notably, Likhitweerawong et al. [16] reported that taurine and homoarginine were elevated in ASD patients, which is contrary to our results.

In addition to the alterations in amino acid levels, which are evident in individuals with ASD and require further research, a variety of lipid abnormalities were detected [51]. In particular, our study highlights the possible causality of ASD-associated changes in fatty acid metabolism. Long-chain polyunsaturated fatty acids (PUFAs) such as arachidonic and docosahexaenoic acids are known to influence neurological responses and behavior [16, 52, 53, 54]. Our research indicates that docosatrienoate (22:3n3), a biomarker of exposure to the polychlorinated biphenyl (PCB) pollutants [55], and the relatively uncharacterized tetradecadienoate (14:2), both belonging to the category of long-chain PUFAs, exert opposite causal effects on the risks of ASD. Additionally, these risks were associated with genetic signature contributing to increased levels of stearidonate (18:4n3), which is a derivative of \alpha-linolenic acid and also elevated after exposure to PCB and in patients with hypertension [56]. Levels of various PUFAs fluctuate across different cohorts or populations, possibly due to differences in lifestyle habits, geographic location, and other factors [37, 57]. For some long-chain saturated fatty acids, such as margaroylcarnitine, Needham et al. [7] also pointed out that there were significant differences between ASD patients and the control group. Moreover, 4-ethylphenyl sulfate was significantly increased in ASD mice. In the plasma of ASD patients, the levels of saturated fatty acid and long-chain acylcarnitines are generally lower than those in matched controls, which is consistent with our findings [7]. These findings suggest that dysregulation in fatty acid metabolism may play a role in the development of ASD [58, 59].

Another important class of metabolites with levels elevated in patients with ASD is derived from steroids, such as androgens and pregnenolone, and these levels correlate with the severity of ASD [59, 60]. In our study, genetic signatures promoting an increase in the levels of two androgenic steroids, 5alpha-androstan-3alpha,17beta-diol disulfate and 11beta-hydroxyetiocholanolone glucuronide, were protective against the risk of ASD. In trials of antioxidant therapy for ASD children, an increase in the levels of certain androgens has been associated with behavioral improvements [61]. Future research needs to further determine the relationship between changes in the levels of individual species of polyunsaturated and steroid lipids and ASD.

The identified metabolites can be regulated through pharmacological interventions or modifiable factors. Specifically, drugs used to treat conditions such as acid sphingomyelinase deficiency or polycystic ovary syndrome can regulate metabolites (sphingomyelin and inositol). In addition, the use of carotene diol or glycerol may be effective in improving symptoms in ASD patients.

Our study has several limitations. MR analysis may introduce biases due to various effects, so we employed multiple models to test the hypotheses. We did not correct for multiple testing on each p-value, which may have increased the risk of false positives. Additionally, some results suggest the presence of potential horizontal pleiotropy, which requires further validation. Our analysis relied solely on genetic factors, so caution is needed when interpreting the results. Furthermore, our data predominantly came from European populations, which may limit the generalizability of the findings. While our study identified certain metabolites associated with ASD risk, this primarily offers predictive insights that require validation. Future research should comprehensively explore the causal relationships and potential molecular mechanisms.

5. Conclusions

Our study suggests that certain plasma metabolites contribute to the risk of ASD, while ASD may affect the composition of the metabolites. These findings suggest the potential for developing new diagnostic tools and therapeutic targets to optimize ASD diagnosis and management.

Availability of Data and Materials

All data generated or analyzed during this study are included in this published article and its supplementary information files.

References

[1] Hirota T, King BH. Autism Spectrum Disorder: A Review. JAMA. 2023; 329: 157–168. https://doi.org/10.1001/jama.2022.23661.

[2] Sharma SR, Gonda X, Tarazi FI. Autism Spectrum Disorder: Classification, diagnosis and therapy. Pharmacology & Therapeutics. 2018; 190: 91–104. https://doi.org/10.1016/j.pharmthera.2018.05.007.

[3] Lyall K, Croen L, Daniels J, Fallin MD, Ladd-Acosta C, Lee BK, et al. The Changing Epidemiology of Autism Spectrum Disorders. Annual Review of Public Health. 2017; 38: 81–102. https://doi.org/10.1146/annurev-publhealth-031816-044318.

[4] Hossain MM, Khan N, Sultana A, Ma P, McKyer ELJ, Ahmed HU, et al. Prevalence of comorbid psychiatric disorders among people with autism spectrum disorder: An umbrella review of systematic reviews and meta-analyses. Psychiatry Research. 2020; 287: 112922. https://doi.org/10.1016/j.psychres.2020.112922.

[5] Rossignol DA, Frye RE. Mitochondrial dysfunction in autism spectrum disorders: a systematic review and meta-analysis. Molecular Psychiatry. 2012; 17: 290–314. https://doi.org/10.1038/mp.2010.136.

[6] El-Ansary A, Bjørklund G, Chirumbolo S, Alnakhli OM. Predictive value of selected biomarkers related to metabolism and oxidative stress in children with autism spectrum disorder. Metabolic Brain Disease. 2017; 32: 1209–1221. https://doi.org/10.1007/s11011-017-0029-x.

[7] Needham BD, Adame MD, Serena G, Rose DR, Preston GM, Conrad MC, et al. Plasma and Fecal Metabolite Profiles in Autism Spectrum Disorder. Biological Psychiatry. 2021; 89: 451–462. https://doi.org/10.1016/j.biopsych.2020.09.025.

[8] Bar N, Korem T, Weissbrod O, Zeevi D, Rothschild D, Leviatan S, et al. A reference map of potential determinants for the human serum metabolome. Nature. 2020; 588: 135–140. https://doi.org/10.1038/s41586-020-2896-2.

[9] Bujak R, Struck-Lewicka W, Markuszewski MJ, Kaliszan R. Metabolomics for laboratory diagnostics. Journal of Pharmaceutical and Biomedical Analysis. 2015; 113: 108–120. https://doi.org/10.1016/j.jpba.2014.12.017.

[10] Wishart DS. Emerging applications of metabolomics in drug discovery and precision medicine. Nature Reviews. Drug Discovery. 2016; 15: 473–484. https://doi.org/10.1038/nrd.2016.32.

[11] Vázquez-Fernández A, Lana A, Struijk EA, Vega-Cabello V, Cárdenas-Valladolid J, Salinero-Fort MÁ et al. Cross-sectional Association Between Plasma Biomarkers and Multimorbidity Patterns in Older Adults. The Journals of Gerontology. Series A. 2024; 79: glad249. https://doi.org/10.1093/gerona/glad249.

[12] Marx W, McGuinness AJ, Rocks T, Ruusunen A, Cleminson J, Walker AJ, et al. The kynurenine pathway in major depressive disorder, bipolar disorder, and schizophrenia: a meta-analysis of 101 studies. Molecular Psychiatry. 2021; 26: 4158–4178. https://doi.org/10.1038/s41380-020-00951-9.

[13] Zhang X, Hu W, Wang Y, Wang W, Liao H, Zhang X, et al. Plasma metabolomic profiles of dementia: a prospective study of 110,655 participants in the UK Biobank. BMC Medicine. 2022; 20: 252. https://doi.org/10.1186/s12916-022-02449-3.

[14] Bot M, Milaneschi Y, Al-Shehri T, Amin N, Garmaeva S, Onderwater GLJ, et al. Metabolomics Profile in Depression: A Pooled Analysis of 230 Metabolic Markers in 5283 Cases With Depression and 10,145 Controls. Biological Psychiatry. 2020; 87: 409–418. https://doi.org/10.1016/j.biopsych.2019.08.016.

[15] Gevi F, Zolla L, Gabriele S, Persico AM. Urinary metabolomics of young Italian autistic children supports abnormal tryptophan and purine metabolism. Molecular Autism. 2016; 7: 47. https://doi.org/10.1186/s13229-016-0109-5.

[16] Likhitweerawong N, Thonusin C, Boonchooduang N, Louthrenoo O, Nookaew I, Chattipakorn N, et al. Profiles of urine and blood metabolomics in autism spectrum disorders. Metabolic Brain Disease. 2021; 36: 1641–1671. https://doi.org/10.1007/s11011-021-00788-3.

[17] Long T, Hicks M, Yu HC, Biggs WH, Kirkness EF, Menni C, et al. Whole-genome sequencing identifies common-to-rare variants associated with human blood metabolites. Nature Genetics. 2017; 49: 568–578. https://doi.org/10.1038/ng.3809.

[18] Hagenbeek FA, Pool R, van Dongen J, Draisma HHM, Jan Hottenga J, Willemsen G, et al. Heritability estimates for 361 blood metabolites across 40 genome-wide association studies. Nature Communications. 2020; 11: 39. https://doi.org/10.1038/s41467-019-13770-6.

[19] Lawlor DA, Harbord RM, Sterne JAC, Timpson N, Davey Smith G. Mendelian randomization: using genes as instruments for making causal inferences in epidemiology. Statistics in Medicine. 2008; 27: 1133–1163. https://doi.org/10.1002/sim.3034.

[20] Lawlor DA, Windmeijer F, Smith GD. Is Mendelian randomization ‘lost in translation?’: comments on ‘Mendelian randomization equals instrumental variable analysis with genetic instruments’ by Wehby et al. Statistics in Medicine. 2008; 27: 2750–2755. https://doi.org/10.1002/sim.3308.

[21] Baranova A, Chandhoke V, Cao H, Zhang F. Shared genetics and bidirectional causal relationships between type 2 diabetes and attention-deficit/hyperactivity disorder. General Psychiatry. 2023; 36: e100996. https://doi.org/10.1136/gpsych-2022-100996.

[22] Baranova A, Cao H, Zhang F. Exploring the influences of education, intelligence and income on mental disorders. General Psychiatry. 2024; 37: e101080. https://doi.org/10.1136/gpsych-2023-101080.

[23] Zhao Q, Baranova A, Cao H, Zhang F. Evaluating Causal Effects of Gut Microbiome on Alzheimer’s Disease. The Journal of Prevention of Alzheimer’s Disease. 2024; 11: 1843–1848. https://doi.org/10.14283/jpad.2024.113.

[24] Liu D, Baranova A, Zhang F. Evaluating the Causal Effect of Type 2 Diabetes on Alzheimer’s Disease Using Large-Scale Genetic Data. The Journal of Prevention of Alzheimer’s Disease. 2024; 11: 1280–1282. https://doi.org/10.14283/jpad.2024.148.

[25] Sun W, Baranova A, Liu D, Cao H, Zhang X, Zhang F. Phenome-wide investigation of bidirectional causal relationships between major depressive disorder and common human diseases. Translational Psychiatry. 2024; 14: 506. https://doi.org/10.1038/s41398-024-03216-z.

[26] Grove J, Ripke S, Als TD, Mattheisen M, Walters RK, Won H, et al. Identification of common genetic risk variants for autism spectrum disorder. Nature Genetics. 2019; 51: 431–444. https://doi.org/10.1038/s41588-019-0344-8.

[27] Chen Y, Lu T, Pettersson-Kymmer U, Stewart ID, Butler-Laporte G, Nakanishi T, et al. Genomic atlas of the plasma metabolome prioritizes metabolites implicated in human diseases. Nature Genetics. 2023; 55: 44–53. https://doi.org/10.1038/s41588-022-01270-1.

[28] 1000 Genomes Project Consortium, Abecasis GR, Auton A, Brooks LD, DePristo MA, Durbin RM, et al. An integrated map of genetic variation from 1,092 human genomes. Nature. 2012; 491: 56–65. https://doi.org/10.1038/nature11632.

[29] Hemani G, Zheng J, Elsworth B, Wade KH, Haberland V, Baird D, et al. The MR-Base platform supports systematic causal inference across the human phenome. Elife. 2018; 7: e34408. https://doi.org/10.7554/eLife.34408.

[30] Wishart DS, Feunang YD, Guo AC, Lo EJ, Marcu A, Grant JR, et al. DrugBank 5.0: a major update to the DrugBank database for 2018. Nucleic Acids Research. 2018; 46: D1074–D1082. https://doi.org/10.1093/nar/gkx1037.

[31] Mendez D, Gaulton A, Bento AP, Chambers J, De Veij M, Félix E, et al. ChEMBL: towards direct deposition of bioassay data. Nucleic Acids Research. 2019; 47: D930–D940. https://doi.org/10.1093/nar/gky1075.

[32] Al-Beltagi M, Saeed NK, Bediwy AS, Elbeltagi R. Metabolomic changes in children with autism. World Journal of Clinical Pediatrics. 2024; 13: 92737. https://doi.org/10.5409/wjcp.v13.i2.92737.

[33] Lingampelly SS, Naviaux JC, Heuer LS, Monk JM, Li K, Wang L, et al. Metabolic network analysis of pre-ASD newborns and 5-year-old children with autism spectrum disorder. Communications Biology. 2024; 7: 536. https://doi.org/10.1038/s42003-024-06102-y.

[34] Liu J, Tan Y, Zhang F, Wang Y, Chen S, Zhang N, et al. Metabolomic analysis of plasma biomarkers in children with autism spectrum disorders. MedComm. 2024; 5: e488. https://doi.org/10.1002/mco2.488.

[35] El-Ansary A, Chirumbolo S, Bhat RS, Dadar M, Ibrahim EM, Bjørklund G. The Role of Lipidomics in Autism Spectrum Disorder. Molecular Diagnosis & Therapy. 2020; 24: 31–48. https://doi.org/10.1007/s40291-019-00430-0.

[36] James SJ, Cutler P, Melnyk S, Jernigan S, Janak L, Gaylor DW, et al. Metabolic biomarkers of increased oxidative stress and impaired methylation capacity in children with autism. The American Journal of Clinical Nutrition. 2004; 80: 1611–1617. https://doi.org/10.1093/ajcn/80.6.1611.

[37] Pastural E, Ritchie S, Lu Y, Jin W, Kavianpour A, Khine Su-Myat K, et al. Novel plasma phospholipid biomarkers of autism: mitochondrial dysfunction as a putative causative mechanism. Prostaglandins, Leukotrienes, and Essential Fatty Acids. 2009; 81: 253–264. https://doi.org/10.1016/j.plefa.2009.06.003.

[38] Ming X, Stein TP, Barnes V, Rhodes N, Guo L. Metabolic perturbance in autism spectrum disorders: a metabolomics study. Journal of Proteome Research. 2012; 11: 5856–5862. https://doi.org/10.1021/pr300910n.

[39] Cheng N, Rho JM, Masino SA. Metabolic Dysfunction Underlying Autism Spectrum Disorder and Potential Treatment Approaches. Frontiers in Molecular Neuroscience. 2017; 10: 34. https://doi.org/10.3389/fnmol.2017.00034.

[40] Sotelo-Orozco J, Abbeduto L, Hertz-Picciotto I, Slupsky CM. Association Between Plasma Metabolites and Psychometric Scores Among Children With Developmental Disabilities: Investigating Sex-Differences. Frontiers in Psychiatry. 2020; 11: 579538. https://doi.org/10.3389/fpsyt.2020.579538.

[41] Rose S, Melnyk S, Pavliv O, Bai S, Nick TG, Frye RE, et al. Evidence of oxidative damage and inflammation associated with low glutathione redox status in the autism brain. Translational Psychiatry. 2012; 2: e134. https://doi.org/10.1038/tp.2012.61.

[42] Díaz-Hung ML, Yglesias-Rivera A, Hernández-Zimbrón LF, Orozco-Suárez S, Ruiz-Fuentes JL, Díaz-García A, et al. Transient glutathione depletion in the substantia nigra compacta is associated with neuroinflammation in rats. Neuroscience. 2016; 335: 207–220. https://doi.org/10.1016/j.neuroscience.2016.08.023.

[43] Gu F, Chauhan V, Chauhan A. Impaired synthesis and antioxidant defense of glutathione in the cerebellum of autistic subjects: alterations in the activities and protein expression of glutathione-related enzymes. Free Radical Biology & Medicine. 2013; 65: 488–496. https://doi.org/10.1016/j.freeradbiomed.2013.07.021.

[44] Tărlungeanu DC, Deliu E, Dotter CP, Kara M, Janiesch PC, Scalise M, et al. Impaired Amino Acid Transport at the Blood Brain Barrier Is a Cause of Autism Spectrum Disorder. Cell. 2016; 167: 1481–1494.e18. https://doi.org/10.1016/j.cell.2016.11.013.

[45] Knaus LS, Basilico B, Malzl D, Gerykova Bujalkova M, Smogavec M, Schwarz LA, et al. Large neutral amino acid levels tune perinatal neuronal excitability and survival. Cell. 2023; 186: 1950–1967.e25. https://doi.org/10.1016/j.cell.2023.02.037.

[46] Nadal-Desbarats L, Aïdoud N, Emond P, Blasco H, Filipiak I, Sarda P, et al. Combined 1H-NMR and 1H-13C HSQC-NMR to improve urinary screening in autism spectrum disorders. The Analyst. 2014; 139: 3460–3468. https://doi.org/10.1039/c4an00552j.

[47] Wang J, Cao Y, Hou W, Bi D, Yin F, Gao Y, et al. Fecal microbiota transplantation improves VPA-induced ASD mice by modulating the serotonergic and glutamatergic synapse signaling pathways. Translational Psychiatry, 2023; 13: 17. https://doi.org/10.1038/s41398-023-02307-7.

[48] Ramírez-Guerrero S, Guardo-Maya S, Medina-Rincón GJ, Orrego-González EE, Cabezas-Pérez R, González-Reyes RE. Taurine and Astrocytes: A Homeostatic and Neuroprotective Relationship. Frontiers in Molecular Neuroscience. 2022; 15: 937789. https://doi.org/10.3389/fnmol.2022.937789.

[49] Ripps H, Shen W. Review: taurine: a “very essential” amino acid. Molecular Vision. 2012; 18: 2673–2686.

[50] Dickinson A, Jones M, Milne E. Measuring neural excitation and inhibition in autism: Different approaches, different findings and different interpretations. Brain Research. 2016; 1648: 277–289. https://doi.org/10.1016/j.brainres.2016.07.011.

[51] Dan Z, Mao X, Liu Q, Guo M, Zhuang Y, Liu Z, et al. Altered gut microbial profile is associated with abnormal metabolism activity of Autism Spectrum Disorder. Gut Microbes. 2020; 11: 1246–1267. https://doi.org/10.1080/19490976.2020.1747329.

[52] Mazahery H, Conlon CA, Beck KL, Mugridge O, Kruger MC, Stonehouse W, et al. A randomised controlled trial of vitamin D and omega-3 long chain polyunsaturated fatty acids in the treatment of irritability and hyperactivity among children with autism spectrum disorder. The Journal of Steroid Biochemistry and Molecular Biology. 2019; 187: 9–16. https://doi.org/10.1016/j.jsbmb.2018.10.017.

[53] Mazahery H, Stonehouse W, Delshad M, Kruger MC, Conlon CA, Beck KL, et al. Relationship between Long Chain n-3 Polyunsaturated Fatty Acids and Autism Spectrum Disorder: Systematic Review and Meta-Analysis of Case-Control and Randomised Controlled Trials. Nutrients. 2017; 9: 155. https://doi.org/10.3390/nu9020155.

[54] Schuchardt JP, Huss M, Stauss-Grabo M, Hahn A. Significance of long-chain polyunsaturated fatty acids (PUFAs) for the development and behaviour of children. European Journal of Pediatrics. 2010; 169: 149–164. https://doi.org/10.1007/s00431-009-1035-8.

[55] Mesnage R, Biserni M, Balu S, Frainay C, Poupin N, Jourdan F, et al. Integrated transcriptomics and metabolomics reveal signatures of lipid metabolism dysregulation in HepaRG liver cells exposed to PCB 126. Archives of Toxicology. 2018; 92: 2533–2547. https://doi.org/10.1007/s00204-018-2235-7.

[56] Al Ashmar S, Anwardeen NR, Anlar GG, Pedersen S, Elrayess MA, Zeidan A. Metabolomic profiling reveals key metabolites associated with hypertension progression. Frontiers in Cardiovascular Medicine. 2024; 11: 1284114. https://doi.org/10.3389/fcvm.2024.1284114.

[57] El-Ansary AK, Bacha AGB, Al-Ayahdi LY. Impaired plasma phospholipids and relative amounts of essential polyunsaturated fatty acids in autistic patients from Saudi Arabia. Lipids in Health and Disease. 2011; 10: 63. https://doi.org/10.1186/1476-511X-10-63.

[58] Esposito CM, Buoli M, Ciappolino V, Agostoni C, Brambilla P. The Role of Cholesterol and Fatty Acids in the Etiology and Diagnosis of Autism Spectrum Disorders. International Journal of Molecular Sciences. 2021; 22: 3550. https://doi.org/10.3390/ijms22073550.

[59] Gillberg C, Fernell E, Kočovská E, Minnis H, Bourgeron T, Thompson L, et al. The role of cholesterol metabolism and various steroid abnormalities in autism spectrum disorders: A hypothesis paper. Autism Research: Official Journal of the International Society for Autism Research. 2017; 10: 1022–1044. https://doi.org/10.1002/aur.1777.

[60] Majewska MD, Hill M, Urbanowicz E, Rok-Bujko P, Bieńkowski P, Namysłowska I, et al. Marked elevation of adrenal steroids, especially androgens, in saliva of prepubertal autistic children. European Child & Adolescent Psychiatry. 2014; 23: 485–498. https://doi.org/10.1007/s00787-013-0472-0.

[61] Bent S, Lawton B, Warren T, Widjaja F, Dang K, Fahey JW, et al. Identification of urinary metabolites that correlate with clinical improvements in children with autism treated with sulforaphane from broccoli. Molecular Autism. 2018; 9: 35. https://doi.org/10.1186/s13229-018-0218-4.