The Metabolic Signature of Autism Spectrum Disorders Using Dried Blood Spots at Birth

Article information

Psychiatry Investig. 2025;22(6):678-686

Publication date (electronic) : 2025 June 16

doi : https://doi.org/10.30773/pi.2024.0293

Haixin Li, Yuqi Yang, Bin Yu, Yuping Zhang

Department of Child Healthcare, Changzhou Maternal and Child Health Care Hospital, Changzhou, China

Correspondence: Yuping Zhang, MM Department of Child Healthcare, Changzhou Maternal and Child Health Care Hospital, Changzhou 213000, China Tel: +86-138-1355-6560, E-mail: yuping6733123@163.com

Correspondence: Bin Yu, MM Department of Child Healthcare, Changzhou Maternal and Child Health Care Hospital, Changzhou 213000, China Tel: +86-138-6125-1515, E-mail: binyu@njmu.edu.cn

Received 2024 September 23; Revised 2025 March 9; Accepted 2025 April 6.

Abstract

INTRODUCTION

Autism spectrum disorder (ASD) is a neurodevelopmental disorder (NDD) that is a collection of early social intercourse dysfunction and restrictive and repetitive behavior patterns, interests, activities and sensory processing anomalies [1]. The prevalence of ASD has been steadily increasing in recent decades which is likely associated with changes in diagnostic criteria, improved performance of screening and diagnostic tools, and increased public awareness [2-4]. The recent overall ASD prevalence was 27.6 per 1,000 (one in 36) children aged 8 years and was 3.8 times as prevalent among boys as among girls based on estimates from the Centers for Disease Control and Prevention’s Autism and Developmental Disabilities Monitoring Network [5]. More than 70% of individuals with autism have concurrent developmental conditions, including intellectual disability (ID), language disorders (LD), and attention-deficit/hyperactivity disorder, as well as medical or psychiatric conditions [6,7].

The etiology of ASD is not fully understood, both genetics and environmental factors early in development play a vital role, among it the genetic basis of the disease accounts for about 50% of cases with ASDs, twin studies have provided strong proof of the heritability of ASD, which is estimated to be even as high as 85% [8-10]. Although its exact cause is not known, several factors have been implicated in its etiology, including inborn errors of metabolism (IEMs), defined by a vast array of disorders that are caused by specific enzyme deficiencies or transport protein defects, is as frequent as in 1 in 800 births [11]. IEMs are particularly frequent as diseases of the nervous system, a review reported that clinical symptoms deriving from central nervous system occur in more than 50% of patients with IEMs [12]. The abnormal functioning of molecules involved in IEM can disrupt neurodevelopment at different stages producing a wide repertoire of clinical manifestations that oscillate from severe brain malformations to mild neurological signs such as ASD relating to amino acids (AAs) [13,14] and carnitine [15-17]. Meanwhile, it has been recently proposed that a significant number of ASD cases could be associated with various metabolic abnormalities, some of them identifiable through untargeted metabolomic profiling [18], simultaneously opening additional space for therapeutic attempts [19].

Recent reports emphasized the causal role of IEMs in ASD, underlying autistic symptoms in less than 5% of cases. The literature on the association between ASD and respiratory chain abnormalities is growing, including complex III/IV deficiency and mitochondrial encephalopathy, lactic acidosis, and stroke-like episodes syndrome, as well as glucose-6-phosphate dehydrogenase deficiency [20]. The discovery of this association and the potential opportunity for therapeutic intervention have prompted the renewed search for characteristic neurometabolic biomarkers to aid in early screening and diagnosis of patients. However, previous studies identified metabolic abnormalities (e.g., AA imbalances, carnitine deficiencies) in ASD children at diagnosis [21], which were confounded by postnatal factors like diet or medication. Neonatal metabolic profiling to distinguish inborn vs. acquired metabolic changes was needed. Our core hypothesis proposes that comprehensive metabolic profiling of a long-term stored dried blood spot (DBS) collected at birth was different in ASDs, enabling us to explore early metabolic signatures while minimizing confounding postnatal influences. Such insight may lead to increased understanding of the etiology, as well as provide a theoretical basis for the early detection and early intervention of ASD.

METHODS

RESULTS

Demographic and metabolism characteristics of ASD and controls divided by sex

As it was shown in Supplementary Table 1, ASD patients (80.2% male vs. 19.8% female) had disproportion of male and female compared to controls (49.1% male vs. 50.9% female) (p<0.001). There were no significant differences in other demographic and metabolism data between ASD patients and controls. Based on their sex, participants were further divided in two groups: male (ASD, n=82; control, n=197) or female (ASD, n=21; control, n=204). Demographic data and statistically significant metabolites in ASD participants with respect to controls divided in sex are shown in Table 1. After adjusted by Benjamini-Hochberg, two carnitines, including C16:1 OH and C18:2, were significantly different in ASD patients compared to sex-matched controls. Univariate logistic regression analyses found no differences between ASD and controls except the sex (Table 2). The forest plot analysis revealed distinct associations between genetic metabolic substances and autism in male and female populations. Among males, the variable C0 demonstrated a statistically significant association (OR=1.05, 95% CI: 1.006–1.096, p=0.024), suggesting a marginal increase in odds per unit change. For females, a significant association was observed with C3 (OR=2.541, 95% CI: 1.089–6.140, p=0.032), indicating a pronounced elevation in odds (Figure 1).

Table 1.

Demographic and metabolism characteristics of ASD patients and controls divided by sex (N=507)

Table 2.

Multivariate regression analysis of the associations between genetic metabolic substances and autism

Figure 1.

The vertical line (OR=1) represents the null effect; confidence intervals crossing this line are non-significant. *p<0.05. ASD, autism spectrum disorder; OR, odds ratio; CI, confidence interval.

Metabolism characteristics of ASD divided by co-occurring conditions

Based on their diagnosis, ASD patients were divided in three groups: 46.22% (49/106) of autistic individuals without any other NDDs were classified into group ASD simply, 30.19% (32/106) of that co-occur with GDD, as well as 23.58% (25/106) with LD. The results showed that autistic individuals co-occurring LD were diagnosed earliest, and the diagnostic age for ASD without NDDs was the largest (p=0.005). No significant differences were found in other demographic features and metabolites (p>0.05) (Table 3). We further divided the three co-occurring groups by sex, showing that in the group of ASD with GDD there were four carnitines, including C3, C4, C5, and C6DC were significantly different between male and female (Supplementary Table 2).

Table 3.

Demographic and metabolism characteristics of autistic individuals divided by co-occurring conditions

Metabolic profile for participants divided by age

Based on their chronological ages of the first diagnosis, autistic individuals were divided in two groups: younger (n=59) or older than 3 years (n=47). Demographic data and metabolic profile of all participants in the two age groups are presented in Table 4. After adjusted by Benjamini-Hochberg, the metabolism of succinic acid (SA) is increased (p<0.05), as well as carnitines C5:1. In younger group, the metabolism of C14OH is significantly different in three groups divided by co-occurring conditions (p=0.049) (Supplementary Table 3).

Table 4.

Metabolic profiles for participants divided by the age of first diagnosis for ASD

DISCUSSION

Features of autism may be detected in early childhood, but the diagnosis of autism is usually not made until much later. Current research has not identified clear biomarkers of autism that can provide a basis for diagnostic criteria, and at this stage, metabolomics has been used to characterize specific perturbations of the metabolome in children with autism compared to neurotypical controls, including AAs, acylcarnitines, and various aromatic and phenolic compounds [22,23]. Given that the metabolome of different compounds, subtle changes identified in a clinical condition can also lead to clues to its mechanism or even treatment, but it’s hard to distinguish IEMs as a primary or secondary pathology [24] because of most of existing studies reflecting cross-sectional differences between autism cases and controls. In order to identify early pathophysiological mechanisms of autism, we analyzed the DBS of future ASDs and controls on birth.

Recent reports find that disorders of IEMs caused by mutations in genes coding for enzymes and other important components in genes coding for enzymes and other important components of various metabolic pathways which are known to be associated with ASD, emphasizing the causal role of IEMs in autism [25,26]. However, in our research, few differences in the neonatal metabolomic profiles were found in ASDs at birth, which was opposite to the metabolic abnormities in autistic children at the diagnostic ages in previous researches. It is well-known that metabolic profiles are affected by several factors, such as diet and gut bacteria, which differ between ASD and typically developing populations and can influence each other. Diet is a major source of circulating metabolites, impacting the metabolome directly or indirectly through chemical transformation by the trillions of gut microbes, the microbiome, which has been proposed to modulate complex behaviors [23]. A recent study supported a model whereby ASD-related restricted interests are associated with less-diverse diet, and in turn reduced microbial taxonomic diversity and looser stool consistency [27]. Our results have implications for the interpretation of cause and effect in relation to the analyses of metabolic profiling in ASDs. There is growing interest in the signature of metabolism to ASDs, but our results emphasize the need to consider the (arguably more intuitive) impact of environmental factors after birth.

As ASD is a NDD with a male/female ratio of 4:1, we considered possible interactions of measured metabolites with sex. We split the sample in two age categories (male and female) to understand possible predictive metabolic signatures capable of distinguishing ASD and controls. The present study identified sex-specific associations between metabolic markers and ASD risk, with elevated C0 (free carnitine) levels linked to ASD in males and higher C3 (propionylcarnitine) levels associated with ASD in females. These findings align with emerging evidence highlighting sexually dimorphic metabolic pathways in NDDs. In males, the role of C0 as a risk factor may reflect disruptions in mitochondrial fatty acid oxidation (FAO), a pathway critical for neurodevelopment. Carnitine deficiency has previously been implicated in ASD pathophysiology, as carnitine facilitates the transport of long-chain fatty acids into mitochondria for energy production [28]. Another study identified mutations in carnitine biosynthesis genes (e.g., SLC6A14) in ASD patients, suggesting impaired carnitine metabolism could contribute to neuronal dysfunction [29]. The male-specific association observed here might be amplified by testosterone’s regulatory effects on carnitine palmitoyltransferase activity, potentially exacerbating metabolic vulnerabilities in males [30].

Conversely, the female-specific link between C3 and ASD may relate to propionate metabolism. Elevated C3 levels indicate propionyl-CoA accumulation, which disrupts mitochondrial function and neurotransmitter synthesis. Propionate overload has been shown to induce ASD-like behaviors in animal models [31], and sex differences in propionate detoxification pathways (e.g., estrogen-enhanced urea cycle activity) might explain the female-specific risk [32]. Additionally, epigenetic modifications in ACADS (acyl-CoA dehydrogenase gene), which are influenced by X-chromosome inactivation patterns, could further drive sex-biased propionylcarnitine accumulation [33].

These results underscore the importance of sex-stratified analyses in ASD research. Future studies should explore hormonal interactions with metabolic pathways and validate these biomarkers in longitudinal cohorts [27].

Autism has historically been considered an impairing condition. Based on the current DSM-5 criteria, ASD includes three levels of severity ranging from “requiring support” to “requiring very substantial support.” Judgments of severity are based solely on the characteristics of the two core domains that make up the diagnostic criteria. However, DSM-5 levels are not sufficient to closely map on to an individual’s real-life functioning, including their experiences and challenges, Waizbard-Bartov et al. [34] pointed that common, co-occurring conditions such as ID, and language delays are as impairing to functioning and wellbeing for many ASD individuals as are the core symptoms themselves, when we analyze the metabolism characteristics of ASDs, we split them in three categories. Although no significant differences were found between the three categories, the American Academy of Pediatrics recommended in favor of an initial metabolic workup in patients with GDD, as these conditions are thought to be more likely of a metabolic nature. This dichotomous approach has raised some concerns including difficulties differentiating the early signs of an isolated ASD from GDD and the paucity of medical providers comfortable with recognizing the signs of an underlying IEM [19].

We also considered possible interactions of measured metabolites with participant ages because of the developmental nature for ASD. Profiles of carnitine and acyl-carnitines change significantly during the first year of life, but kept at the same level between 2 and 15 years [35]. We split the sample in two age categories (<3 and ≥3 years) to understand possible predictive metabolic signatures capable of distinguishing ASD and individuals at early stages of ASD development. This threshold is consistent with reliable ASD diagnosis and effectiveness of early intervention. Indeed, the definite diagnosis of ASD is generally made between 3 and 5 years [36]. Moreover, increasing evidences support the effectiveness of early interventions (behavioral, developmental, and educational approaches) in pre-schoolers (aged 24–71 months) with ASD [36]. In our study, the results showed that compared with autistic children younger than 3 years of age, the concentrations of succinylacetone were higher in ASDs older than 3 years of age at diagnosis (p=0.013), and C5:1 also differed between the two ages of diagnosis (p=0.006). The observed age-related differences in SA and carnitine C5:1 metabolism between children diagnosed with ASD before and after age 3 years may reflect distinct developmental trajectories in mitochondrial and lipid metabolic pathways [37,38]. The observation that SA and carnitine C5:1 levels measured at birth differ between children later diagnosed with ASD before and after age 3 years suggests that early metabolic dysregulation may contribute to heterogeneous ASD trajectories. Elevated neonatal SA levels in children diagnosed after age 3 years could reflect congenital mitochondrial dysfunction or disrupted TCA cycle flux, predisposing to progressive neurodevelopmental challenges. Rossignol and Frye [37] hypothesized that mitochondrial defects in ASD may originate prenatally, with SA accumulation acting as a compensatory response to impaired oxidative phosphorylation. This aligns with studies linking neonatal metabolic markers to later ASD risk [38]. Lower neonatal C5:1 levels in children diagnosed <3 years may indicate inborn errors of FAO or carnitine shuttle defects, disrupting energy supply during critical neurodevelopmental windows. However, postnatal environmental factors (e.g., gut microbiota maturation) might modulate these neonatal metabolic risks, explaining later diagnosis in some cases [39]. These findings underscore the potential of neonatal metabolic profiling to stratify ASD subtypes and guide early interventions. Future studies should validate whether SA/C5:1 at birth predict ASD severity or developmental trajectories, while controlling for confounders like maternal metabolic health or perinatal stressors.

Our research, taking into account factors such as sex, comorbidities, and the age of diagnosis, found a few significant differences in the metabolic spectra of those born with ASD. However, our study had several limitations including a small sample size, blood samples collected at the time of diagnosis for new screening data analysis and less stringent inclusion criteria for controls (e.g., healthy controls matched by sex and birth date without NDDs or inherited metabolic diseases). Besides, the metabolic signature of ASDs at diagnostic ages is not considered in this study. Finally, correlation analyses between metabolite levels and clinical scores were not conducted due to the absence of granular clinical data (e.g., Autism Diagnostic Observation Schedule scores) in our cohort. In the future, more samples and more detailed study designs are needed in our study.

In conclusion, relatively little is known about the metabolic impact of the ASD, while such insight may lead to increased understanding of the etiology. We analyzed the metabolic nature at birth, finding that most analytes included in the TMS screen had no significant differences between the autism group and the control group at birth; however, sex, the age of first diagnose for ASD, and comorbidities may be the important factors affecting metabolic characteristics, as well as the genetic metabolic changes arise after birth.

Supplementary Materials

Article information Continued